CA2048882C - Image data processing method and apparatus - Google Patents
Image data processing method and apparatusInfo
- Publication number
- CA2048882C CA2048882C CA 2048882 CA2048882A CA2048882C CA 2048882 C CA2048882 C CA 2048882C CA 2048882 CA2048882 CA 2048882 CA 2048882 A CA2048882 A CA 2048882A CA 2048882 C CA2048882 C CA 2048882C
- Authority
- CA
- Canada
- Prior art keywords
- image
- coefficients
- transform coefficients
- dimensional
- quantized
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/24—Systems for the transmission of television signals using pulse code modulation
- H04N7/52—Systems for transmission of a pulse code modulated video signal with one or more other pulse code modulated signals, e.g. an audio signal or a synchronizing signal
- H04N7/54—Systems for transmission of a pulse code modulated video signal with one or more other pulse code modulated signals, e.g. an audio signal or a synchronizing signal the signals being synchronous
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/103—Selection of coding mode or of prediction mode
- H04N19/112—Selection of coding mode or of prediction mode according to a given display mode, e.g. for interlaced or progressive display mode
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
- H04N19/126—Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/157—Assigned coding mode, i.e. the coding mode being predefined or preselected to be further used for selection of another element or parameter
- H04N19/16—Assigned coding mode, i.e. the coding mode being predefined or preselected to be further used for selection of another element or parameter for a given display mode, e.g. for interlaced or progressive display mode
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/44—Decoders specially adapted therefor, e.g. video decoders which are asymmetric with respect to the encoder
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
- H04N19/93—Run-length coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/30—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
Abstract
An image data reconstructing method and apparatus, adaptive to both progressive and sequential image buildup modes, reconstructs an orthogonally transformed image by dequantizing transform coefficients decoded from coded data of a continuous tone image, by causing the transform coefficients to have a common data structure through an attachment of a dummy signal representing a zero value DC element.
The image data reconstructing method comprises a coded data decoding step, a common data structure one-dimensional quantized transform coefficient creating step, a one-to-two dimensional quantized transform coefficient transforming step, a dequantization threshold storing step, and a two-dimensional quantized transform coefficient dequantizing step. The image data reconstructing apparatus comprises a dummy signal generator for generating such dummy signals and a selector for selecting the dummy signals.
The image data reconstructing method comprises a coded data decoding step, a common data structure one-dimensional quantized transform coefficient creating step, a one-to-two dimensional quantized transform coefficient transforming step, a dequantization threshold storing step, and a two-dimensional quantized transform coefficient dequantizing step. The image data reconstructing apparatus comprises a dummy signal generator for generating such dummy signals and a selector for selecting the dummy signals.
Description
Image Data Processlng Method and Apparatus Background of the Inventlon This inventlon relates to an lmage data reconstructlng method and apparatus for reconstructing an lmage from coded data of a continuous tone lmage. More partlculary lt relates to an lmage data reconstructlng apparatus and method for reconstructlng by selectlng sequentlal lmage bulldup or progresslve lmage bulldup from coded data of a contlnuous tone lmage coded by the adaptlve dlscrete coslne transform codlng.
Brief Descrlption of the Drawings Figure 1 is a block diagram of a coding circuit per the ADCT method, Flgure 2 is a table of exemplary data for an orlglnal lmage recelved by a conventlonal two-dlmenslonal DCT
unlt;
Flgure 3 ls a table of exemplary DCT coefficients representlng space frequency dlstrlbutlons;
Flgure 4 ls a block diagram of a conventlonal two-dlmenslonal DCT unlt, Flgure 5 ls a table of thresholds for quantlzlng DCT
coefflclents, Figure 6 is a table of quantized DCT coefflclents;
Flgure 7 ls a dlagram of the order of zlgzag scannlng of the generated quantlzed coefflclents;
Flgure 8 ls a block dlagram of a conventlonal decodlng clrcult per the ADCT method;
Flgure 9 ls a block diagram of a conventional two-*
`- 2o48882 dimensional inverse DCT unit of a ADCT decodlng clrcult;
Flgure 10 ls a block dlagram of a conventional dequantizer;
Figures llA, llB and llC are diagrams of the conventional data structures of one-dimenslonal quantlzed DCT
coefflclents for use ln a sequentlal lmage bulldup, the flrst stage of a progresslve lmage bulldup and the second and later stages of a progressive image bulldup, respectlvely;
Flgure 12 ls a dlagram of an example of dlvldlng two~dlmenslonal quantlzed DCT coefflclents lnto bands, ie one-dimensional quantlzed DCT coefflclent reglons, ln a progresslve lmage bulldup, Flgures 13A, 13B and 13C are dlagrams of the data structures of one-dlmenslonal quantlzed DCT coefflclents for use ln a sequentlal image bulldup, the flrst state of a progressive lmage bulldup and the second and later stages of a progressive image bulldup, respectlvely;
Flgure 14 ls a block dlagram of an embodlment of thls invention;
Flgure 15 ls a flowchart showlng the control loglc of a dequantlzatlon controller for use in a progresslve lmage bulldup, and Flgure 16 ls a block dlagram of an lmage data reconstructlng clrcult for reconstructlng lmage data from the decoded dequantlzed DCT coefficlents outputted from the dequantizer shown ln Figure 14.
Description of the Related Art Conventionally, an adaptlve dlscrete cosine ~ transform codlng rnethod is used as a hlghly efflclent method 2a . 28151-43 20488~2 for compressing lmage data, for example.
The following ls a descrlptlon of the adaptlve dlscrete coslne transform coding method (, or the ADCT method ln short).
Accordlng to the ADCT method, images are dlvlded lnto blocks comprlslng e.g. 8X8 plcture elements. DCT
coefflclents expresslng the dlstrlbutlon of space frequencles are determlned by performlng two-dlmenslonal dlscrete cosine transforms (hereafter ,.~
abbreviated as DCTs ) for the image signals from respective divided blocks. The image signals are coded by quantizing the DCT coefficients using thresholds corresponding to visual requirements, according to a Huffman table obtained statistically from the quantized coefficients.
Figure 1 is a block diagram of a conventional coding circuit per the ADCT method.
Figure 2 shows exemplary data for an original image received by a conventional two-dimensional DCT
unit.
Figure 3 shows exemplary DCT coefficients representing space frequency distributions.
The coding operation per the ADCT method is described below.
A two-dimensional DCT unit 24 receives from an input terminal 23 sixty-four (64) image signals, such as those shown in Figure 2, for a block comprising 8X8 picture elements. By orthogonally transforming the received image signals, the two-dimensional DCT unit 24 transforms them into coefficients having space frequency distributions such as those shown in Figure 3, thereby calculating DCT coefficients, and outputs them to a linear quantizer 25.
Figure 4 is a block diagram of the two-20~8882 dimensional DCT unit 24. A one-dimensional DCT unit 30 performs one-dimensional DCTs for the image signals inputted from the input terminal 23, and a transposer 31 transposes the matrix obtained from the one-5 dimensional DCT unit 30. Then, a one-dimensional DCT unit 32 performs one dimensional DCTs, as with the one-dimensional DCT unit 30 described earlier. As with the transposer 31 described earlier, a transposer 33 transposes the matrix obtained at the one-10 dimensional DCT unit 32 and outputs it to a terminal34. Through similar performances for all blocks of image blocks, the inputted image signals are transformed to the DCT coefficients.
Figure 5 is a table of thresholds for quantizing 15 DCT coefficients.
Figure 6 is a table of exemplary quantized DCT
coefficients.
Further explanation is made by referring back to Figure 1. The linear quantizer 25 quantizes the 20 inputted DCT coefficients by the quantization matrix 22 comprising the thresholds shown in Figure 5 and determined based on visual experimentation. The linear quantization produces quantized coefficients such as those shown in Figure 6, which indicate that 25 the DCT coefficients smaller than thresholds become 0, 20~82 thereby forming quantized coefficients in which the DC
elements and only a few AC elements have non-zero values.
Generally, the higher the visual sensitivities the lower the space frequencies are, and the lower the visual sensitivities the higher the space frequencies are. Therefore, as shown in Figure 5, the absolute values of the quantization matrix elements, i.e. the thresholds for DCT coefficients corresponding to lower space frequencies, are smaller, and the absolute values corresponding to higher space frequencies are larger. Accordingly, as shown in Figure 6, of quantized coefficients, only the DC element at the top left end and a very few AC elements representing low space frequency elements become significant coefficients having non-zero values, whereas all others become insignificant coefficients, in most cases.
Figure 7 shows the order of scanning the generated quantized coefficients (from the lower frequency elements to the higher frequency elements of the space frequencies).
The linear quantizer 25 transforms the two-dimensionally arrayed quantized coefficients into a one-dimensional form according to the zigzag scanning order shown ln Figure 7 and outputs them to a variable length coder 26 in the next stage.
Upon receipt of these quantized coefficients, the variable length coder 26 codes the quantlzed 5 coefficlents by referring to a coding table 27 composed of a Huffman table created by statlstical volumes for the respective images. As for their DC elements, the differences between the DC elements of the present and preceding blocks are coded in variable lengths. As for their AC elements, the values of the AC
elements (hereafter referred to as indlces~ of signlflcant coefficients (non-zero value coefficients) and the run lengths ~hereafter referred to as runs) of the insignificant coefficients (zero value coefficients) are coded in variable lengths. The output terminal 28 sequentially outputs the coded data to the external units.
Meanwhile, the coded data obtained by the coding circuit per the ADCT method are reconstructed as images according to the following method.
Figure 8 is a block diagram of a conventional decoding circuit per the ADCT method.
A variable length decoder 41 receives the coded data inputted from an input terminal 40. The variable length decoder 41 decodes the received coded , ~
20~8~8~2 data into fixed length data of the indices and runs and outputs the decoded data to a dequantizer 43, according to a decoding table 42 formed by a table which is the inverse of the Huffman table composing the coding table 27.
On receiving the decoded data (the decoded quantized coefficients), the dequantizer 43 restores the dequantized DCT coefficients through a dequantization by multiplying the respective decoded data by the thresholds stored at the corresponding positions in a quantization matrix 48. The dequantizer 43 outputs the restored dequantized DCT
coefficients to a two-dimensional inverse DCT unit 44.
The two-dimensional inverse DCT unit 44 orthogonally transforms the received dequantized DCT
coefficients into image signals indicating the distribution of the space frequencies.
The two-dimensional inverse DCT unit 44 is explained in further detail.
Figure 9 is a block diagram of a conventional two-dimensional inverse DCT unit of an ADCT decoding circuit.
A one-dimensional inverse dequantized DCT unit 51 performs one-dimensional inverse DCTs for the DCT
coefficients inputted from a terminal 50 and outputs 20~8882 ,9 ~
them to a transposer 52. The transposer 52 transposes the matrix representing the outputs from the one-dimensional inverse DCT unit 51. A one-dimensional inverse DCT unit 53 again performs one-dimensional inverse DCTs on the transposed matrix obtained from the matrix transposition at the transposer 52. As with the transposer 52, a transposer 54 transposes the matrix representing the outputs from the one-dimensional inverse DCT unit 53.
A terminal 45 displays signals obtained by these processes, thereby reconstructing the images.
A conventional method for reconstructing coded data transformed by ADCT is the sequential image buildup which is the same as the sequential reconstruction method used for a generic hard copy communication in which an image is reconstructed from the top to the bottom.
However, since the sequential reconstruction method takes a constant amount of time for reconstruction regardless of the data volume, a progressive image buildup method is adopted for high-speed searching such as searching through a database, by which a rough image is reconstructed at an early stage, however, the image quality is bad, and the image quality is improved gradually.
20488~2 Figure 10 ls a block dlagram of a conventlonal dequantlzer.
Slnce an ADCT needs to deal wlth two types of reconstructlng methods whlch are sequentlal lmage bulldup and progresslve lmage bulldup, a conventlonal dequantlzer 43 ls conflgured as shown ln Flgure 10.
In Flgure 10, a varlable length decoder 41 decodes coded data lnputted from an lnput termlnal 60, and a sequentlal lrnage bulldup coefflcient reconstructor 62 and a progresslve lmage bulldup coefflclent reconstructor 63 recelve the decoded one-dlmenslonal quantlzed DCT coefflclents, the dequantlzatlon of whlch ls controlled by a dequantlzatlon controller 66, accordlng to a slgnal STA for speclfylng the klnd of reconstructlng method lnputted from an lnput termlnal 61.
Flgures llA, llB and llC respectlvely show the data structures of conventlonal one-dlmenslonal quantlzed DCT
coefflclents for use ln a sequentlal lmage bulldup, the flrst stage of a progresslve lmage bulldup and the second and later stages of a progresslve lmage bulldup.
As shown ln Flgure llA the one-dlmenslonal quantlzed DCT coefficlents recelved by the sequential image buildup coefflclent reconstructor 62 have a conflguratlon such that the head end of a block ls always a DC element coefflcient D5 and the rest comprlses runs (R0 and R5) and lndices (I-2, I-3,...). When the signal STA specifies sequentlal reconstructlon, two-dlmenslonal quantlzed DCT coefflclents ln all reglons ln a block are reconstructed from respective , 281Sl-43 20488û2 one-dimenslonal DCT coefflclents, selected by a selector 64 such as a multlplexer, and recelved and stored by a quantlzed coefflclent storlng unlt 65.
A multlplier 105 dequantizes quantized DCT
coefficients stored in the quantized coefficient storing unit 65 by sequentially multiplying them by quantization thresholds stored ln the quantlzatlon threshold storing unit 2, which are outputted from an output termlnal 67.
Flgure 12 shows an example of dividing two-dimensional quantlzed DCT coefflclents lnto bands, l.e.one-dlmensional quantlzed DCT coefficlent regions, ln a progresslve lmage bulldup.
As shown ln Figure 12, a block is divlded into plural kinds of regions (hereafter referred to as "bands"), and the coefflcients in a band are selected as signlficant coefflcients, and all others are replaced by "0" in respectlve stages of a progresslve ~, 204~8~2 image buildup by ADCT .
More concretely, when band 1 (having only a DC
element) is selected in a first stage, since the inputted one-dimensional quantized DCT coefficients comprise only coefficients of the DC element, as shown in Figure 11B, and two-dimensional quantized DCT
coefficients are reconstructed by setting the rest of the 8X8 two dimensional quantized DCT coefficients obtained by coding the 8X8 picture elements of a block in an image, i.e. the sixty-three (63) AC element coefficients, to "O".
When band 2 is selected in a second stage, since one-dimensional quantized DCT coefficients comprise runs indicating the number of zeros and indices indicating the values of AC elements, as shown in Figure 11C, coefficients in band 2 is reconstructed from respective coefficient values and coefficients in all other regions are set to "O", thereby reconstructing two-dimensional quantized DCT
coefficients.
Thereafter, similarly, coefficients in bands corresponding to respective stages are reconstructed as significant coefficients, and coefficients in all other regions are set to "O", thereby reconstructing two-dimensional quantized DCT coefficients.
-The progressive coefficient reconstructor 63 performs the following operations for a progressive image buildup. When the dequantization controller 66 selects a progressive image buildup, a coefficient 5 reconstructor 71 in the progressive coefficient reconstructor 63 receives one-dimensional quantized DCT coefficients according to the stage number PST of the specified progressive image buildup, and reconstructs two-dimensional quantized DCT
10 coefficients according to methods corresponding to respective stages. An address generator 72 generates two-dimensional addresses for the head end and tail end of the respective reconstruction stages.
A mask processor 73 masks coefficients in regions 15 other than the generated addresses to "0". The selector 64 selects two-dimensional quantized DCT
coefficients in all regions in a block. The quantized coefficient storing unit 65 receives the two-dimensional quantized DCT coefficients selected by 20 the selector 64. That is, the quantized coefficient storing unit 65 stores the two-dimensional quantized DCT coef ficients outputted from the progressive coefficient reconstructor 63.
The multiplier 105 sequentially reads the two-25 dimensional quantized coefficients stored in the 204~882 1~
quantized coefficient storing unit 65 and multiplies them by quantization thresholds stored in the quantization threshold storing unit 2, thereby dequantizing them, and outputting the result to the terminal 67.
Thus, a dequantization according to either sequential image buildup or progressive image buildup is performed.
However, a conventional image data reconstructing method or apparatus for performing both sequential image buildup and progressive image buildup such as the one described above has a problem of the circuit size being larger than the circuit size of a circuit corresponding to only the image buildup method, because it requires separate dequantizing circuits for sequential image buildup and progressive image buildup, controls operations of the circuit corresponding to the selected reconstruction method, and selects obtained DCT coefficients.
Summary of the Invention This invention relates to an image data reconstructing method and apparatus for reconstructing an image from coded data of a continuous tone image.
More particularly, it relates to an image data ,~
reconstructing apparatus and method for reconstruction by selecting sequential image buildup or progressive image buildup, from coded data of a continuous tone image, coded by the adaptive discrete cosine transform 5 coding method.
It aims at downsizing the image data reconstructing circuit by standardizing the data structure of the quantized DCT coefficients in a sequential image buildup with that of the data 10 structure in a progressive image buildup.
Hence, when the coded data do not have the DC
element dif ference between the current block and the preceding block, this invention causes dummy signals indicating zero value DC elements to be attached to the head ends of data structures of quantized DCT
coefficients having only AC elements in the second and later stages of a progressive image buildup, so that such quantized DCT coef ficients have the same data structure as the quantized DCT coef ficients in the 20 first stage of a progressive image buildup having the DC element.
A feature of this invention resides in an image data reconstructing method comprising a coded data decoding step, a common data structure one-dimensional 25 quantized DCT coefficient creating step, a one-to-two . 2o48882 dlmenslonal quantlzed DCT coefflclent transformlng step, a dequantlzatlon threshold storlng step, and a two-dlmenslonal quantlzed DCT coefficient dequantlzlng step.
Another feature of thls lnventlon resldes ln an lmage data reconstructing apparatus embodying the above method. The crucial distlnctlon of such an apparatus lncludes a dummy slgnal generator for generatlng such dummy slgnals and a selector for selectlng the dummy slgnals. Slnce the data structure of the quantlzed DCT coefflclents ln the flrst stage has the DC element at the head end and an end-of-block slgnal REOB at the tall end as wlth the data structure of quantlzed DCT coefflclents ln a sequentlal lmage bulldup, a common dequantlzer adaptively performs a sequential image bulldup and a progresslve lmage bulldup. Such dummy slgnals the selector selects can be elther dummy coded data or the dummy decoded data belng zero value DCT coefflclents.
In accordance wlth the present lnventlon there ls provlded an lmage data reconstructlng apparatus, for reconstructlng an lmage from coded data obtalned by codlng quantlzed coefflclents, whlch are obtalned by quantlzlng transform coefflclents obtalned by orthogonally transformlng gradlent values of plural plcture elements ln respectlve blocks each comprlsing a plurality of plcture elements dlvlded from an orlglnal lmage, for progresslvely bulldlng up flne lmages from rough lmages ln the progresslve lmage bulldup mode; sald lmage data reconstructlng apparatus comprlslng, a dummy data generator for generatlng dummy data; a selector for selectlng elther sald dummy data or sald coded data; and a controller for having sald selector select only said coded data ln a first stage, and said dummy data and then sald coded data ln second and later stages.
In accordance wlth the present lnventlon there ls also provided an lmage data reconstructlng method for reconstructlng an lmage from coded data obtalned by codlng quantlzed coefflclents, whlch are obtalned by quantlzlng transform coefflclents obtalned by orthogonally transformlng gradlent values of plural plcture elements ln respectlve blocks each comprislng a plurallty of plcture elements dlvlded from an original lrnage, ln whlch either a sequential image bulldup mode for sequentlally bulldlng up lmages or a progressive image buildup mode for progresslvely bullding up flne images from rough images is selected; said image data reconstructlng method comprising: a first step of decodlng coded data lnto one-dimenslonal quantized transform coefficients; a second step of creating one-dimensional quantized transform coefflclents havlng a common data structure for use by both the sequentlal lmage bulldup mode and the progresslve lmage bulldup mode; a thlrd step of transformlng sald one-dlmensional quantized transform coefflcients created in said second step into two-dlmenslonal quantized transform coefficients; a fourth step of storing thresholds in preparatlon for dequantizing quantized transform coefflclents lnto dequantlzed transform coefficients; and a fifth step of dequantizing said two-dimensional quantized transform coefflcients obtained in said third step by said thresholds stored in sald fourth step; wherein: said 16a ;--, 28151-43 ~ . .. .... . .
thresholds stored ln sald fourth step sequentlally dequantlze sald quantlzed transform coefflcients created in sald second and thlrd steps regardless of the lmage buildup mode in said flfth step.
In accordance wlth the present lnvention there ls further provlded an lmage data reconstructing apparatus, for reconstructing an lmage from coded data obtalned by codlng quantlzed coefflcients, whlch are obtained by quantlzlng transform coefflclents obtalned by orthogonally transformlng gradlent values of plural plcture elements ln respectlve blocks each comprlsing a plurallty of plcture elements dlvlded from an orlglnal lmage, ln whlch elther a sequentlal lmage bulldup mode for sequentlally bulldlng up lmages or a progresslve lmage bulldup mode for progresslvely bulldlng up flne lmages from rough lmages ls selected; sald lmage data reconstructlng apparatus comprlslng: a dequantlzatlon controller as an lmage buildup mode specifier for selectlng elther the sequentlal lmage buildup mode for sequentlally reconstructing an image or the progressive lmage bulldup mode for progressively reconstructlng a flne lmage from a rough lmage; a varlable length decoder for decoding coded data lnto one-dlmenslonal quantlzed transform coefflclents; a quantlzed coefflclent creator for creating from one-dimenslonal quantlzed transform coefflcients decoded by sald varlable length decoder one-dlmenslonal quantlzed transform coefflclents havlng a common data structure for both the sequentlal lmage bulldup mode and the progresslve lmage bulldup mode; a quantlzed coefflclent receptor for 16b -; 28151-43 transforming one-dlmenslonal quantlzed transform coefflclents lnto two-dlmenslonal quantized transform coefflclents and for storing the latter; a quantlzation threshold storing unit for storing thresholds for dequantizlng quantlzed transform coefflcients; and a multlplier as a dequantization calculator for multlplylng for dequantlzations two-dlmenslonal quantlzed orthogonal transform coefficlents stored in sald quantlzed coefflclent receptor by sald thresholds stored in said quantization threshold storing unit wherein:
one-dimensional quantlzed transform coefficients having a data structure common to both the sequential image bulldup mode and the progressive lmage bulldup mode are created for a specification of either one.
Description of the Preferred Embodiments Flgures 13A, 13B and 13C respectively show the data structures of one-dimenslonal quantlzed DCT coefficients for use in a sequential image buildup, the first stage of a progressive lmage bulldup and the second and later stages of a progresslve image buildup.
The embodiment relates to an lmage data 16c , .. . . . ..
reconstructing apparatus and method for reconstructlng an lmage from coded data obtained by codlng quantlzed coefflclents, whlch are obtalned by quantlzlng DCT
coefficlents obtalned by performlng two-dlmenslonal dlscrete cosine transforms on the gradlent values of plural plcture elements ln respectlve blocks each comprlslng a plurallty (N x N; where N ls a posltlve lnteger) of plcture elements from an orlglnal lmage, ln whlch elther the sequentlal lmage bulldup mode for sequentlally bulldlng up images or the progresslve lmage bulldup mode for progresslvely bulldlng up flne lmages from rough images ls selected.
Thls lnventlon conflgures such an lmage data reconstructlng method to comprlse a flrst step of decodlng coded data lnto one-dlmenslonal quantized DCT coefflclents, a second step of creatlng one-dlmensional quantlzed DCT
coefflclents havlng a common data structure comprlslng a DC
element coefflclent at the head end as well as a run RM
speclfylng the number of zeros and an lndex I speclfylng non-zero AC elements for use by both the sequentlal lmage bulldup mode (Flgure 13A) and the progresslve lmage bulldup mode (Flgures 13B and 13C), a thlrd step of transformlng the one-dlmenslonal quantlzed DCT coefflclents created ln the second step lnto two-C
20~888~
dimensional quantized DCT coefficients, a fourth step of storing thresholds for dequantizing quantized DCT
coefficients into dequantized DCT coefficients, and a fifth step of dequantizing the two-dimensional quantized DCT coefficients obtained in the third step by applying the thresholds stored in the fourth step, wherein the thresholds stored in the fourth step sequentially dequantize the quantized DCT coefficients created in the second and third steps regardless of the image buildup mode in the fifth step.
Here, the second step also adds dummy signals to the head ends of the one-dimensional quantized DCT
coefficients for the second and later stages decoded in the first step during a progressive image buildup.
The dummy signals are DCT coefficients having zero values.
Figure 14 is a block diagram of an embodiment of this invention.
This invention configures an image data reconstructing apparatus (Figure 14) to comprise a dequantization controller 83 for selecting either the sequential image buildup mode for sequentially reconstructing an image or the progressive image buildup mode for progressively reconstructing a fine image from a rough image, a variable length decoder 80 .
for decoding coded data into one-dimensional quantized DCT coefficients, a quantized coefficient creator 82 for creating from one-dimensional quantized DCT
coefficients decoded by the variable length decoder 80 one-dimensional quantized DCT coefficients having a common data structure for both the sequential image buildup mode and the progressive image buildup mode, a quantized coefficient receptor 84 for transforming one-dimensional quantized DCT coefficients into two-dimensional quantized DCT coefficients and for storingthe latter, a quantization threshold storing unit 85 for storing thresholds for dequantizing quantized DCT
coefficients, and a multiplier 86 for dequantization by multiplying two-dimensional quantized DCT
coefficients stored in the quantized coefficient receptor 84 by the thresholds stored in the quantization threshold storing unit 85, wherein one-dimensional quantized DCT coefficients having a data structure common to both the sequential image buildup mode and the progressive image buildup mode are created for a specification of either one.
Here, the quantized coefficient creator 82 has a dummy signal generator 88 for generating dummy signals, which are DCT coefficients having zero values, to be attached to the head ends of the one-dimensional quantized DCT coefficients decoded by thevariable length decoder 80 during a progressive image buildup.
Also, the quantized coefficient creator 82 has a selector 81, which may be a multiplexer, for selecting the one-dimensional quantized DCT coefficients from the variable length decoder 80 and dummy signals from the dummy signal generator 88. When the dequantization controller 83 specifies a progressive image buildup, the selector 81 attaches dummy signals to the head ends of one-dimensional quantized DCT
coefficients outputted from the quantized coefficient creator 82 in the second and later stages.
The image data reconstructing method and apparatus of this invention configured as above has a common dequantizing circuit to adaptively dequantize two-dimensional quantized DCT coefficients obtained by unifying the data structures of decoded one-dimensional quantized DCT coefficients in both the sequential image buildup mode and the progressive image buildup mode, thereby eliminating the necessity for having separate circuits respectively for the either mode.
The following is a further explanation of the image data reconstructing method and apparatus of this --` 2o48882 lnvention by referring to Figure 14.
The dequantizing circuit shown in Flgure 14 is used when coded data obtalned by an ADCT coding circuit as shown ln Flgures 1 and 2 ls used for an lmage reconstructlon. The two-dimenslonal lnverse DCT unlt 44 shown ln Figure 8 receives the outputs from the dequantizing circuit lllustrated ln Flgure 14.
In Flgure 14, the varlable length decoder 80 decodes coded data lnputted from an lnput termlnal 87 to one-dlmensional quantized DCT coefflclents and outputs them to the selector 81. The dummy slgnal generator 88 constantly outputs to the selector 81 dummy slgnals, whlch become a dummy slgnal "D0" lndlcatlng that the DC elements at the head ends of one-dlmenslonal quantlzed DCT coefflclents have zero values. The dequantlzatlon controller 83 as a reconstructlon mode specifier outputs a selectlon slgnal SEL to the selector 81 enabling the selector 81 to selectively output slgnals from the variable length decoder 80 and the dummy signal generator 88. The dequantlzatlon controller 83 receives a specification signal STA specifying either the sequentlal lmage bulldup mode or the progresslve lmage bulldup mode from an lnput termlnal 89.
When the sequentlal lmage bulldup mode is speclfled, the dequantlzatlon controller, 83 lnstructs the selector 81 to constantly select decoded slgnals from the varlable length decoder 80.
The quantlzed coefficient receptor 84 sequentlally recelves one-dlmenslonal quantlzed DCT coefflclents decoded ~' 2 0 ~
from coded data by the varlable length decoder 80 and selected by the selector 81 and reconstructs them to two-dlmenslonal quantlzed DCT coefflclents ln a predetermlned manner.
That ls, one-dlmenslonal quantlzed DCT coefflclents outputted from the varlable length decoder 80 such as shown ln Flgure 13A are transformed to two-dlmenslonal quantlzed DCT
coefflclents in the followlng manner. As for the DC element ln Flgure 13A slnce the value DL at the head end represents DC
element codes in a variable length indicating the difference from the DC element ln the precedlng block, the DC element coefflcient is obtalned by addlng the dlfference to the DC
element reconstructed ln the precedlng block. As for AC
elements, slnce the run RM lndicates the number of zeros ln successlon, the AC element coefficients are stored next as indices Il, I2, .... IN.
~ .s a result, the quantized coefficient receptor 84 has a circuit configuratlon equivalent to combining both the sequential coefficient reconstructor 62 and the quantization coefficlent storlng unlt 65 lllustrated ln prlor art Flgure 10.
The multlplier 86 reads two-dimensional quantized DCT coefflcients for a block stored in the quantized coefflclent receptor 84 to be multlplled by the quantlzatlon thresholds stored ln the quantlzatlon threshold storlng unit 85 for dequantization and then outputted from an output termlnal ~1.
On the other hand, when the progressive image bulldup mode ls speclfled, slnce the data structure of j,.-. ~
~ 28151-43 one-dimensional quantized DCT coefficients decoded by the variable length decoder 80 for the first stage is such that the head end has a value DL specifying the DC element, as shown ln Flgure 13B, the data structure is the same as that for the sequential image bulldup mode shown ln Figure 13A.
The selector 81 selects outputs from the variable length decoder 80 and supplies them to the quantized coefficient receptor 84 for dequantization with the same processing as that for the sequential image buildup mode.
As illustrated in Flgure llC, ln and after the second stage, the data structure of the one-dimensional quantlzed DCT coefflcients decoded by the variable length decoder 80 does not have the value indicating a DC element at the head end and is - 204~882 different from that for the sequential image buildup mode. Therefore, this invention causes the dequantization controller 83 in the second and later stages to instruct the selector 81 to select dummy signals outputted from the dummy signal generator 88 at the timings for reading head ends and to output the dummy signals to the quantized coefficient receptor 84. This processing makes the data structure of the one-dimensional quantized DCT coefficients, as shown in Figure 13C, which the quantized coefficient receptor 84 receives to be such that the head end becomes a DC element as with the data structure shown in Figure 13A for the sequential image buildup mode, so that one-dimensional quantized DCT coefficients have the data structure enabling them to be dequantized in the same manner whether the image reconstruction is in the sequential image buildup mode or in the progressive image buildup mode.
By repeating the dequantization for a whole image in block units, dequantized DCT coefficients are reconstructed from quantized coefficients for the whole image.
Figure 15 is a flowchart showing the controls by the dequantization controller 83 described earlier.
After an image buildup starts, step S1 begins.
Step S1: The dequantization controller 83 determines whether the image reconstruction is in the sequential image buildup mode or in the progressive image buildup mode, upon receiving the specification signal STA.
When the image reconstruction is determined to be in the sequential image buildup mode, the control proceeds to step S2.
When the image reconstruction is determined to be in the progressive image buildup mode, the control proceeds to step S3.
Step S2: Since the image data is reconstructed in a single pass, the dequantization controller 83 outputs to the selector 81 a selection signal SEL instructing the selector 81 to select one-dimensional quantized DCT coefficients decoded by the variable length decoder 80. This causes the output terminal 91 to output two-dimensional dequantized DCT coefficients in the sequential image buildup mode.
Step S3: It is judged whether or not the progressive image buildup is in the first stage.
If the decision is affirmative, the control proceeds to step S2. Since the data structure in the first stage of the progressive image buildup, e.g.
having only a DC element and others being all zero, is the same as that of the sequential image buildup, step S2 is executed.
If the decision is negative, the control proceeds ~8~2 to step S4.
Step S4: The dequantlzation controller 83 outputs the selectlon slgnal SEL to the selector 81 lnstructlng lt to select the dummy slgnal "D0" generated by the dummy signal generator 88. The selector 81 ln turn outputs the dummy slgnal "D0" to the quantlzed coefflclent receptor 84. Step S5:
Then, the dequantlzatlon controller 83 outputs the selectlon slgnal SEL to the selector 81 lnstructlng lt to select outputs from the varlable length decoder 80. Thls change ln selectlon causes the selector 81 to output signals having the data structure shown ln Flgure 13C.
A repetition of the above steps completes the controls for processlng a whole lmage ln the sequentlal lmage bulldup mode or each stage ln the progresslve bulldup mode untll a detailed whole lmage ls obtalned.
Regardless of the stage of the progresslve lmage bulldup, the data structure of the slgnals, outputted from the selector 81 under control to the dequantlzatlon controller 83, comprlses the dlfference from the precedlng block or lts dummy slgnal "D0" at the head end, runs RM and indlces IN. When an image is reconstructed from the data structure, conventionally, the klnds of the data structures need to be declded flrst and dequantlzatlon needs to be performed accordlng to the lmage buildup modes. However, slnce this lnventlon conflgures the same clrcult to dequantize decoded DCT coefficients, it elimlnates the necesslty for such a decision.
In a progresslve image buildup, image data are reconstructed ln stage unlts from DCT coefflclents outputted from the multlpller 86 reconstructed also ln stage unlts.
After the flrst stage, the newly reconstructed lmage data are added to the hltherto reconstructed lmage data to obtaln lmage data more closely representlng the orlglnal lmage.
As descrlbed above, thls lnventlon enables a slngle common clrcult to adaptlvely dequantlze decoded one-dlmenslonal quantlzed DCT coefflclents havlng a data structure correspondlng to elther the sequentlal lmage bulldup mode or the progresslve lmage bulldup mode, thereby enabllng the clrcult to be downsized.
Flgure 16 ls a block dlagram of an lmage data reconstructlng clrcult for reconstructlng lmage data from the decoded dequantlzed DCT coefflclents outputted from the dequantlzer shown ln Flgure 14. A system controller 95 outputs to the dequantlzatlon controller 83 the speclflcatlon slgnal STA speclfylng the sequential lmage bulldup or the progresslve lmage bulldup, and controls an lmage storlng unlt 96 and an adder 97.
The two-dlmenslonal lnverse DCT unlt 44 performs lnverse DCTs on decoded dequantlzed DCT coefflclents supplled through the output termlnal 91 lnto reconstructed lmage data, whlch have two klnds. Image data DA will be used to refer to one klnd of data for the sequentlal image bulldup and the progresslve lmage buildup ln the flrst stage. Image data DB
wlll be used to refer to the other klnd of data for the progresslve lmage bulldup ln the second and later stages.
The system controller 95 constantly keeps track of e 28151 43 whlch klnd of lmage data ls outputted from the two-dlmensional lnverse DCT unlt 44.
In the case of lmage data DA, the system controller 95 halts the operatlons of the adder 97, and causes the lmage storlng unlt 96 to recelve and store outputs from the two-dlmenslonal lnverse DCT unlt 44 "as ls" at the storage addresses the system controller 95 speclfles.
In the case of lmage data DB, the system controller 95 actlvates the operatlons of the adder 97, so that the newly lnputted lmage data DB ls added to the correspondlng lmage data hltherto accumulated ln the lmage storlng unit 96, then stores the result J
in the same image storing unit 96.
The adder 97 adds image data received after stage 1 in a progressive image buildup to the image data hitherto accumulated in the image storing unit 96 for fine-tuning the reconstructed image.
Figure 16 shows a configuration in which the system controller 95 halts the operations of the adder 97 and causes the image storing unit 96 to store the outputs from the two-dimensional inverse DCT unit 44 "as is". However, an embodiment of this invention can adopt a configuration such that the system controller 95 clears the image data stored in the image storing unit 96.
Although the image data reconstructing method and apparatus described in the above embodiments assume dummy signals or decoded data are selected, it is possible configure them to select dummy signals or coded data before being decoded.
Although the image data reconstructing method and apparatus described in the above embodiments assume image processing through two-dimensional discrete cosine transforms, this invention is also applicable to image processing through other methods of orthogonal transforms.
Brief Descrlption of the Drawings Figure 1 is a block diagram of a coding circuit per the ADCT method, Flgure 2 is a table of exemplary data for an orlglnal lmage recelved by a conventlonal two-dlmenslonal DCT
unlt;
Flgure 3 ls a table of exemplary DCT coefficients representlng space frequency dlstrlbutlons;
Flgure 4 ls a block diagram of a conventlonal two-dlmenslonal DCT unlt, Flgure 5 ls a table of thresholds for quantlzlng DCT
coefflclents, Figure 6 is a table of quantized DCT coefflclents;
Flgure 7 ls a dlagram of the order of zlgzag scannlng of the generated quantlzed coefflclents;
Flgure 8 ls a block dlagram of a conventlonal decodlng clrcult per the ADCT method;
Flgure 9 ls a block diagram of a conventional two-*
`- 2o48882 dimensional inverse DCT unit of a ADCT decodlng clrcult;
Flgure 10 ls a block dlagram of a conventional dequantizer;
Figures llA, llB and llC are diagrams of the conventional data structures of one-dimenslonal quantlzed DCT
coefflclents for use ln a sequentlal lmage bulldup, the flrst stage of a progresslve lmage bulldup and the second and later stages of a progressive image bulldup, respectlvely;
Flgure 12 ls a dlagram of an example of dlvldlng two~dlmenslonal quantlzed DCT coefflclents lnto bands, ie one-dimensional quantlzed DCT coefflclent reglons, ln a progresslve lmage bulldup, Flgures 13A, 13B and 13C are dlagrams of the data structures of one-dlmenslonal quantlzed DCT coefflclents for use ln a sequentlal image bulldup, the flrst state of a progressive lmage bulldup and the second and later stages of a progressive image bulldup, respectlvely;
Flgure 14 ls a block dlagram of an embodlment of thls invention;
Flgure 15 ls a flowchart showlng the control loglc of a dequantlzatlon controller for use in a progresslve lmage bulldup, and Flgure 16 ls a block dlagram of an lmage data reconstructlng clrcult for reconstructlng lmage data from the decoded dequantlzed DCT coefficlents outputted from the dequantizer shown ln Figure 14.
Description of the Related Art Conventionally, an adaptlve dlscrete cosine ~ transform codlng rnethod is used as a hlghly efflclent method 2a . 28151-43 20488~2 for compressing lmage data, for example.
The following ls a descrlptlon of the adaptlve dlscrete coslne transform coding method (, or the ADCT method ln short).
Accordlng to the ADCT method, images are dlvlded lnto blocks comprlslng e.g. 8X8 plcture elements. DCT
coefflclents expresslng the dlstrlbutlon of space frequencles are determlned by performlng two-dlmenslonal dlscrete cosine transforms (hereafter ,.~
abbreviated as DCTs ) for the image signals from respective divided blocks. The image signals are coded by quantizing the DCT coefficients using thresholds corresponding to visual requirements, according to a Huffman table obtained statistically from the quantized coefficients.
Figure 1 is a block diagram of a conventional coding circuit per the ADCT method.
Figure 2 shows exemplary data for an original image received by a conventional two-dimensional DCT
unit.
Figure 3 shows exemplary DCT coefficients representing space frequency distributions.
The coding operation per the ADCT method is described below.
A two-dimensional DCT unit 24 receives from an input terminal 23 sixty-four (64) image signals, such as those shown in Figure 2, for a block comprising 8X8 picture elements. By orthogonally transforming the received image signals, the two-dimensional DCT unit 24 transforms them into coefficients having space frequency distributions such as those shown in Figure 3, thereby calculating DCT coefficients, and outputs them to a linear quantizer 25.
Figure 4 is a block diagram of the two-20~8882 dimensional DCT unit 24. A one-dimensional DCT unit 30 performs one-dimensional DCTs for the image signals inputted from the input terminal 23, and a transposer 31 transposes the matrix obtained from the one-5 dimensional DCT unit 30. Then, a one-dimensional DCT unit 32 performs one dimensional DCTs, as with the one-dimensional DCT unit 30 described earlier. As with the transposer 31 described earlier, a transposer 33 transposes the matrix obtained at the one-10 dimensional DCT unit 32 and outputs it to a terminal34. Through similar performances for all blocks of image blocks, the inputted image signals are transformed to the DCT coefficients.
Figure 5 is a table of thresholds for quantizing 15 DCT coefficients.
Figure 6 is a table of exemplary quantized DCT
coefficients.
Further explanation is made by referring back to Figure 1. The linear quantizer 25 quantizes the 20 inputted DCT coefficients by the quantization matrix 22 comprising the thresholds shown in Figure 5 and determined based on visual experimentation. The linear quantization produces quantized coefficients such as those shown in Figure 6, which indicate that 25 the DCT coefficients smaller than thresholds become 0, 20~82 thereby forming quantized coefficients in which the DC
elements and only a few AC elements have non-zero values.
Generally, the higher the visual sensitivities the lower the space frequencies are, and the lower the visual sensitivities the higher the space frequencies are. Therefore, as shown in Figure 5, the absolute values of the quantization matrix elements, i.e. the thresholds for DCT coefficients corresponding to lower space frequencies, are smaller, and the absolute values corresponding to higher space frequencies are larger. Accordingly, as shown in Figure 6, of quantized coefficients, only the DC element at the top left end and a very few AC elements representing low space frequency elements become significant coefficients having non-zero values, whereas all others become insignificant coefficients, in most cases.
Figure 7 shows the order of scanning the generated quantized coefficients (from the lower frequency elements to the higher frequency elements of the space frequencies).
The linear quantizer 25 transforms the two-dimensionally arrayed quantized coefficients into a one-dimensional form according to the zigzag scanning order shown ln Figure 7 and outputs them to a variable length coder 26 in the next stage.
Upon receipt of these quantized coefficients, the variable length coder 26 codes the quantlzed 5 coefficlents by referring to a coding table 27 composed of a Huffman table created by statlstical volumes for the respective images. As for their DC elements, the differences between the DC elements of the present and preceding blocks are coded in variable lengths. As for their AC elements, the values of the AC
elements (hereafter referred to as indlces~ of signlflcant coefficients (non-zero value coefficients) and the run lengths ~hereafter referred to as runs) of the insignificant coefficients (zero value coefficients) are coded in variable lengths. The output terminal 28 sequentially outputs the coded data to the external units.
Meanwhile, the coded data obtained by the coding circuit per the ADCT method are reconstructed as images according to the following method.
Figure 8 is a block diagram of a conventional decoding circuit per the ADCT method.
A variable length decoder 41 receives the coded data inputted from an input terminal 40. The variable length decoder 41 decodes the received coded , ~
20~8~8~2 data into fixed length data of the indices and runs and outputs the decoded data to a dequantizer 43, according to a decoding table 42 formed by a table which is the inverse of the Huffman table composing the coding table 27.
On receiving the decoded data (the decoded quantized coefficients), the dequantizer 43 restores the dequantized DCT coefficients through a dequantization by multiplying the respective decoded data by the thresholds stored at the corresponding positions in a quantization matrix 48. The dequantizer 43 outputs the restored dequantized DCT
coefficients to a two-dimensional inverse DCT unit 44.
The two-dimensional inverse DCT unit 44 orthogonally transforms the received dequantized DCT
coefficients into image signals indicating the distribution of the space frequencies.
The two-dimensional inverse DCT unit 44 is explained in further detail.
Figure 9 is a block diagram of a conventional two-dimensional inverse DCT unit of an ADCT decoding circuit.
A one-dimensional inverse dequantized DCT unit 51 performs one-dimensional inverse DCTs for the DCT
coefficients inputted from a terminal 50 and outputs 20~8882 ,9 ~
them to a transposer 52. The transposer 52 transposes the matrix representing the outputs from the one-dimensional inverse DCT unit 51. A one-dimensional inverse DCT unit 53 again performs one-dimensional inverse DCTs on the transposed matrix obtained from the matrix transposition at the transposer 52. As with the transposer 52, a transposer 54 transposes the matrix representing the outputs from the one-dimensional inverse DCT unit 53.
A terminal 45 displays signals obtained by these processes, thereby reconstructing the images.
A conventional method for reconstructing coded data transformed by ADCT is the sequential image buildup which is the same as the sequential reconstruction method used for a generic hard copy communication in which an image is reconstructed from the top to the bottom.
However, since the sequential reconstruction method takes a constant amount of time for reconstruction regardless of the data volume, a progressive image buildup method is adopted for high-speed searching such as searching through a database, by which a rough image is reconstructed at an early stage, however, the image quality is bad, and the image quality is improved gradually.
20488~2 Figure 10 ls a block dlagram of a conventlonal dequantlzer.
Slnce an ADCT needs to deal wlth two types of reconstructlng methods whlch are sequentlal lmage bulldup and progresslve lmage bulldup, a conventlonal dequantlzer 43 ls conflgured as shown ln Flgure 10.
In Flgure 10, a varlable length decoder 41 decodes coded data lnputted from an lnput termlnal 60, and a sequentlal lrnage bulldup coefflcient reconstructor 62 and a progresslve lmage bulldup coefflclent reconstructor 63 recelve the decoded one-dlmenslonal quantlzed DCT coefflclents, the dequantlzatlon of whlch ls controlled by a dequantlzatlon controller 66, accordlng to a slgnal STA for speclfylng the klnd of reconstructlng method lnputted from an lnput termlnal 61.
Flgures llA, llB and llC respectlvely show the data structures of conventlonal one-dlmenslonal quantlzed DCT
coefflclents for use ln a sequentlal lmage bulldup, the flrst stage of a progresslve lmage bulldup and the second and later stages of a progresslve lmage bulldup.
As shown ln Flgure llA the one-dlmenslonal quantlzed DCT coefficlents recelved by the sequential image buildup coefflclent reconstructor 62 have a conflguratlon such that the head end of a block ls always a DC element coefflcient D5 and the rest comprlses runs (R0 and R5) and lndices (I-2, I-3,...). When the signal STA specifies sequentlal reconstructlon, two-dlmenslonal quantlzed DCT coefflclents ln all reglons ln a block are reconstructed from respective , 281Sl-43 20488û2 one-dimenslonal DCT coefflclents, selected by a selector 64 such as a multlplexer, and recelved and stored by a quantlzed coefflclent storlng unlt 65.
A multlplier 105 dequantizes quantized DCT
coefficients stored in the quantized coefficient storing unit 65 by sequentially multiplying them by quantization thresholds stored ln the quantlzatlon threshold storing unit 2, which are outputted from an output termlnal 67.
Flgure 12 shows an example of dividing two-dimensional quantlzed DCT coefflclents lnto bands, l.e.one-dlmensional quantlzed DCT coefficlent regions, ln a progresslve lmage bulldup.
As shown ln Figure 12, a block is divlded into plural kinds of regions (hereafter referred to as "bands"), and the coefflcients in a band are selected as signlficant coefflcients, and all others are replaced by "0" in respectlve stages of a progresslve ~, 204~8~2 image buildup by ADCT .
More concretely, when band 1 (having only a DC
element) is selected in a first stage, since the inputted one-dimensional quantized DCT coefficients comprise only coefficients of the DC element, as shown in Figure 11B, and two-dimensional quantized DCT
coefficients are reconstructed by setting the rest of the 8X8 two dimensional quantized DCT coefficients obtained by coding the 8X8 picture elements of a block in an image, i.e. the sixty-three (63) AC element coefficients, to "O".
When band 2 is selected in a second stage, since one-dimensional quantized DCT coefficients comprise runs indicating the number of zeros and indices indicating the values of AC elements, as shown in Figure 11C, coefficients in band 2 is reconstructed from respective coefficient values and coefficients in all other regions are set to "O", thereby reconstructing two-dimensional quantized DCT
coefficients.
Thereafter, similarly, coefficients in bands corresponding to respective stages are reconstructed as significant coefficients, and coefficients in all other regions are set to "O", thereby reconstructing two-dimensional quantized DCT coefficients.
-The progressive coefficient reconstructor 63 performs the following operations for a progressive image buildup. When the dequantization controller 66 selects a progressive image buildup, a coefficient 5 reconstructor 71 in the progressive coefficient reconstructor 63 receives one-dimensional quantized DCT coefficients according to the stage number PST of the specified progressive image buildup, and reconstructs two-dimensional quantized DCT
10 coefficients according to methods corresponding to respective stages. An address generator 72 generates two-dimensional addresses for the head end and tail end of the respective reconstruction stages.
A mask processor 73 masks coefficients in regions 15 other than the generated addresses to "0". The selector 64 selects two-dimensional quantized DCT
coefficients in all regions in a block. The quantized coefficient storing unit 65 receives the two-dimensional quantized DCT coefficients selected by 20 the selector 64. That is, the quantized coefficient storing unit 65 stores the two-dimensional quantized DCT coef ficients outputted from the progressive coefficient reconstructor 63.
The multiplier 105 sequentially reads the two-25 dimensional quantized coefficients stored in the 204~882 1~
quantized coefficient storing unit 65 and multiplies them by quantization thresholds stored in the quantization threshold storing unit 2, thereby dequantizing them, and outputting the result to the terminal 67.
Thus, a dequantization according to either sequential image buildup or progressive image buildup is performed.
However, a conventional image data reconstructing method or apparatus for performing both sequential image buildup and progressive image buildup such as the one described above has a problem of the circuit size being larger than the circuit size of a circuit corresponding to only the image buildup method, because it requires separate dequantizing circuits for sequential image buildup and progressive image buildup, controls operations of the circuit corresponding to the selected reconstruction method, and selects obtained DCT coefficients.
Summary of the Invention This invention relates to an image data reconstructing method and apparatus for reconstructing an image from coded data of a continuous tone image.
More particularly, it relates to an image data ,~
reconstructing apparatus and method for reconstruction by selecting sequential image buildup or progressive image buildup, from coded data of a continuous tone image, coded by the adaptive discrete cosine transform 5 coding method.
It aims at downsizing the image data reconstructing circuit by standardizing the data structure of the quantized DCT coefficients in a sequential image buildup with that of the data 10 structure in a progressive image buildup.
Hence, when the coded data do not have the DC
element dif ference between the current block and the preceding block, this invention causes dummy signals indicating zero value DC elements to be attached to the head ends of data structures of quantized DCT
coefficients having only AC elements in the second and later stages of a progressive image buildup, so that such quantized DCT coef ficients have the same data structure as the quantized DCT coef ficients in the 20 first stage of a progressive image buildup having the DC element.
A feature of this invention resides in an image data reconstructing method comprising a coded data decoding step, a common data structure one-dimensional 25 quantized DCT coefficient creating step, a one-to-two . 2o48882 dlmenslonal quantlzed DCT coefflclent transformlng step, a dequantlzatlon threshold storlng step, and a two-dlmenslonal quantlzed DCT coefficient dequantlzlng step.
Another feature of thls lnventlon resldes ln an lmage data reconstructing apparatus embodying the above method. The crucial distlnctlon of such an apparatus lncludes a dummy slgnal generator for generatlng such dummy slgnals and a selector for selectlng the dummy slgnals. Slnce the data structure of the quantlzed DCT coefflclents ln the flrst stage has the DC element at the head end and an end-of-block slgnal REOB at the tall end as wlth the data structure of quantlzed DCT coefflclents ln a sequentlal lmage bulldup, a common dequantlzer adaptively performs a sequential image bulldup and a progresslve lmage bulldup. Such dummy slgnals the selector selects can be elther dummy coded data or the dummy decoded data belng zero value DCT coefflclents.
In accordance wlth the present lnventlon there ls provlded an lmage data reconstructlng apparatus, for reconstructlng an lmage from coded data obtalned by codlng quantlzed coefflclents, whlch are obtalned by quantlzlng transform coefflclents obtalned by orthogonally transformlng gradlent values of plural plcture elements ln respectlve blocks each comprlsing a plurality of plcture elements dlvlded from an orlglnal lmage, for progresslvely bulldlng up flne lmages from rough lmages ln the progresslve lmage bulldup mode; sald lmage data reconstructlng apparatus comprlslng, a dummy data generator for generatlng dummy data; a selector for selectlng elther sald dummy data or sald coded data; and a controller for having sald selector select only said coded data ln a first stage, and said dummy data and then sald coded data ln second and later stages.
In accordance wlth the present lnventlon there ls also provided an lmage data reconstructlng method for reconstructlng an lmage from coded data obtalned by codlng quantlzed coefflclents, whlch are obtalned by quantlzlng transform coefflclents obtalned by orthogonally transformlng gradlent values of plural plcture elements ln respectlve blocks each comprislng a plurallty of plcture elements dlvlded from an original lrnage, ln whlch either a sequential image bulldup mode for sequentlally bulldlng up lmages or a progressive image buildup mode for progresslvely bullding up flne images from rough images is selected; said image data reconstructlng method comprising: a first step of decodlng coded data lnto one-dimenslonal quantized transform coefficients; a second step of creating one-dimensional quantized transform coefflclents havlng a common data structure for use by both the sequentlal lmage bulldup mode and the progresslve lmage bulldup mode; a thlrd step of transformlng sald one-dlmensional quantized transform coefflcients created in said second step into two-dlmenslonal quantized transform coefficients; a fourth step of storing thresholds in preparatlon for dequantizing quantized transform coefflclents lnto dequantlzed transform coefficients; and a fifth step of dequantizing said two-dimensional quantized transform coefflcients obtained in said third step by said thresholds stored in sald fourth step; wherein: said 16a ;--, 28151-43 ~ . .. .... . .
thresholds stored ln sald fourth step sequentlally dequantlze sald quantlzed transform coefflcients created in sald second and thlrd steps regardless of the lmage buildup mode in said flfth step.
In accordance wlth the present lnvention there ls further provlded an lmage data reconstructing apparatus, for reconstructing an lmage from coded data obtalned by codlng quantlzed coefflcients, whlch are obtained by quantlzlng transform coefflclents obtalned by orthogonally transformlng gradlent values of plural plcture elements ln respectlve blocks each comprlsing a plurallty of plcture elements dlvlded from an orlglnal lmage, ln whlch elther a sequentlal lmage bulldup mode for sequentlally bulldlng up lmages or a progresslve lmage bulldup mode for progresslvely bulldlng up flne lmages from rough lmages ls selected; sald lmage data reconstructlng apparatus comprlslng: a dequantlzatlon controller as an lmage buildup mode specifier for selectlng elther the sequentlal lmage buildup mode for sequentlally reconstructing an image or the progressive lmage bulldup mode for progressively reconstructlng a flne lmage from a rough lmage; a varlable length decoder for decoding coded data lnto one-dlmenslonal quantlzed transform coefflclents; a quantlzed coefflclent creator for creating from one-dimenslonal quantlzed transform coefflcients decoded by sald varlable length decoder one-dlmenslonal quantlzed transform coefflclents havlng a common data structure for both the sequentlal lmage bulldup mode and the progresslve lmage bulldup mode; a quantlzed coefflclent receptor for 16b -; 28151-43 transforming one-dlmenslonal quantlzed transform coefflclents lnto two-dlmenslonal quantized transform coefflclents and for storing the latter; a quantlzation threshold storing unit for storing thresholds for dequantizlng quantlzed transform coefflcients; and a multlplier as a dequantization calculator for multlplylng for dequantlzations two-dlmenslonal quantlzed orthogonal transform coefficlents stored in sald quantlzed coefflclent receptor by sald thresholds stored in said quantization threshold storing unit wherein:
one-dimensional quantlzed transform coefficients having a data structure common to both the sequential image bulldup mode and the progressive lmage bulldup mode are created for a specification of either one.
Description of the Preferred Embodiments Flgures 13A, 13B and 13C respectively show the data structures of one-dimenslonal quantlzed DCT coefficients for use in a sequential image buildup, the first stage of a progressive lmage bulldup and the second and later stages of a progresslve image buildup.
The embodiment relates to an lmage data 16c , .. . . . ..
reconstructing apparatus and method for reconstructlng an lmage from coded data obtained by codlng quantlzed coefflclents, whlch are obtalned by quantlzlng DCT
coefficlents obtalned by performlng two-dlmenslonal dlscrete cosine transforms on the gradlent values of plural plcture elements ln respectlve blocks each comprlslng a plurallty (N x N; where N ls a posltlve lnteger) of plcture elements from an orlglnal lmage, ln whlch elther the sequentlal lmage bulldup mode for sequentlally bulldlng up images or the progresslve lmage bulldup mode for progresslvely bulldlng up flne lmages from rough images ls selected.
Thls lnventlon conflgures such an lmage data reconstructlng method to comprlse a flrst step of decodlng coded data lnto one-dlmenslonal quantized DCT coefflclents, a second step of creatlng one-dlmensional quantlzed DCT
coefflclents havlng a common data structure comprlslng a DC
element coefflclent at the head end as well as a run RM
speclfylng the number of zeros and an lndex I speclfylng non-zero AC elements for use by both the sequentlal lmage bulldup mode (Flgure 13A) and the progresslve lmage bulldup mode (Flgures 13B and 13C), a thlrd step of transformlng the one-dlmenslonal quantlzed DCT coefflclents created ln the second step lnto two-C
20~888~
dimensional quantized DCT coefficients, a fourth step of storing thresholds for dequantizing quantized DCT
coefficients into dequantized DCT coefficients, and a fifth step of dequantizing the two-dimensional quantized DCT coefficients obtained in the third step by applying the thresholds stored in the fourth step, wherein the thresholds stored in the fourth step sequentially dequantize the quantized DCT coefficients created in the second and third steps regardless of the image buildup mode in the fifth step.
Here, the second step also adds dummy signals to the head ends of the one-dimensional quantized DCT
coefficients for the second and later stages decoded in the first step during a progressive image buildup.
The dummy signals are DCT coefficients having zero values.
Figure 14 is a block diagram of an embodiment of this invention.
This invention configures an image data reconstructing apparatus (Figure 14) to comprise a dequantization controller 83 for selecting either the sequential image buildup mode for sequentially reconstructing an image or the progressive image buildup mode for progressively reconstructing a fine image from a rough image, a variable length decoder 80 .
for decoding coded data into one-dimensional quantized DCT coefficients, a quantized coefficient creator 82 for creating from one-dimensional quantized DCT
coefficients decoded by the variable length decoder 80 one-dimensional quantized DCT coefficients having a common data structure for both the sequential image buildup mode and the progressive image buildup mode, a quantized coefficient receptor 84 for transforming one-dimensional quantized DCT coefficients into two-dimensional quantized DCT coefficients and for storingthe latter, a quantization threshold storing unit 85 for storing thresholds for dequantizing quantized DCT
coefficients, and a multiplier 86 for dequantization by multiplying two-dimensional quantized DCT
coefficients stored in the quantized coefficient receptor 84 by the thresholds stored in the quantization threshold storing unit 85, wherein one-dimensional quantized DCT coefficients having a data structure common to both the sequential image buildup mode and the progressive image buildup mode are created for a specification of either one.
Here, the quantized coefficient creator 82 has a dummy signal generator 88 for generating dummy signals, which are DCT coefficients having zero values, to be attached to the head ends of the one-dimensional quantized DCT coefficients decoded by thevariable length decoder 80 during a progressive image buildup.
Also, the quantized coefficient creator 82 has a selector 81, which may be a multiplexer, for selecting the one-dimensional quantized DCT coefficients from the variable length decoder 80 and dummy signals from the dummy signal generator 88. When the dequantization controller 83 specifies a progressive image buildup, the selector 81 attaches dummy signals to the head ends of one-dimensional quantized DCT
coefficients outputted from the quantized coefficient creator 82 in the second and later stages.
The image data reconstructing method and apparatus of this invention configured as above has a common dequantizing circuit to adaptively dequantize two-dimensional quantized DCT coefficients obtained by unifying the data structures of decoded one-dimensional quantized DCT coefficients in both the sequential image buildup mode and the progressive image buildup mode, thereby eliminating the necessity for having separate circuits respectively for the either mode.
The following is a further explanation of the image data reconstructing method and apparatus of this --` 2o48882 lnvention by referring to Figure 14.
The dequantizing circuit shown in Flgure 14 is used when coded data obtalned by an ADCT coding circuit as shown ln Flgures 1 and 2 ls used for an lmage reconstructlon. The two-dimenslonal lnverse DCT unlt 44 shown ln Figure 8 receives the outputs from the dequantizing circuit lllustrated ln Flgure 14.
In Flgure 14, the varlable length decoder 80 decodes coded data lnputted from an lnput termlnal 87 to one-dlmensional quantized DCT coefflclents and outputs them to the selector 81. The dummy slgnal generator 88 constantly outputs to the selector 81 dummy slgnals, whlch become a dummy slgnal "D0" lndlcatlng that the DC elements at the head ends of one-dlmenslonal quantlzed DCT coefflclents have zero values. The dequantlzatlon controller 83 as a reconstructlon mode specifier outputs a selectlon slgnal SEL to the selector 81 enabling the selector 81 to selectively output slgnals from the variable length decoder 80 and the dummy signal generator 88. The dequantlzatlon controller 83 receives a specification signal STA specifying either the sequentlal lmage bulldup mode or the progresslve lmage bulldup mode from an lnput termlnal 89.
When the sequentlal lmage bulldup mode is speclfled, the dequantlzatlon controller, 83 lnstructs the selector 81 to constantly select decoded slgnals from the varlable length decoder 80.
The quantlzed coefficient receptor 84 sequentlally recelves one-dlmenslonal quantlzed DCT coefflclents decoded ~' 2 0 ~
from coded data by the varlable length decoder 80 and selected by the selector 81 and reconstructs them to two-dlmenslonal quantlzed DCT coefflclents ln a predetermlned manner.
That ls, one-dlmenslonal quantlzed DCT coefflclents outputted from the varlable length decoder 80 such as shown ln Flgure 13A are transformed to two-dlmenslonal quantlzed DCT
coefflclents in the followlng manner. As for the DC element ln Flgure 13A slnce the value DL at the head end represents DC
element codes in a variable length indicating the difference from the DC element ln the precedlng block, the DC element coefflcient is obtalned by addlng the dlfference to the DC
element reconstructed ln the precedlng block. As for AC
elements, slnce the run RM lndicates the number of zeros ln successlon, the AC element coefficients are stored next as indices Il, I2, .... IN.
~ .s a result, the quantized coefficient receptor 84 has a circuit configuratlon equivalent to combining both the sequential coefficient reconstructor 62 and the quantization coefficlent storlng unlt 65 lllustrated ln prlor art Flgure 10.
The multlplier 86 reads two-dimensional quantized DCT coefflcients for a block stored in the quantized coefflclent receptor 84 to be multlplled by the quantlzatlon thresholds stored ln the quantlzatlon threshold storlng unit 85 for dequantization and then outputted from an output termlnal ~1.
On the other hand, when the progressive image bulldup mode ls speclfled, slnce the data structure of j,.-. ~
~ 28151-43 one-dimensional quantized DCT coefficients decoded by the variable length decoder 80 for the first stage is such that the head end has a value DL specifying the DC element, as shown ln Flgure 13B, the data structure is the same as that for the sequential image bulldup mode shown ln Figure 13A.
The selector 81 selects outputs from the variable length decoder 80 and supplies them to the quantized coefficient receptor 84 for dequantization with the same processing as that for the sequential image buildup mode.
As illustrated in Flgure llC, ln and after the second stage, the data structure of the one-dimensional quantlzed DCT coefflcients decoded by the variable length decoder 80 does not have the value indicating a DC element at the head end and is - 204~882 different from that for the sequential image buildup mode. Therefore, this invention causes the dequantization controller 83 in the second and later stages to instruct the selector 81 to select dummy signals outputted from the dummy signal generator 88 at the timings for reading head ends and to output the dummy signals to the quantized coefficient receptor 84. This processing makes the data structure of the one-dimensional quantized DCT coefficients, as shown in Figure 13C, which the quantized coefficient receptor 84 receives to be such that the head end becomes a DC element as with the data structure shown in Figure 13A for the sequential image buildup mode, so that one-dimensional quantized DCT coefficients have the data structure enabling them to be dequantized in the same manner whether the image reconstruction is in the sequential image buildup mode or in the progressive image buildup mode.
By repeating the dequantization for a whole image in block units, dequantized DCT coefficients are reconstructed from quantized coefficients for the whole image.
Figure 15 is a flowchart showing the controls by the dequantization controller 83 described earlier.
After an image buildup starts, step S1 begins.
Step S1: The dequantization controller 83 determines whether the image reconstruction is in the sequential image buildup mode or in the progressive image buildup mode, upon receiving the specification signal STA.
When the image reconstruction is determined to be in the sequential image buildup mode, the control proceeds to step S2.
When the image reconstruction is determined to be in the progressive image buildup mode, the control proceeds to step S3.
Step S2: Since the image data is reconstructed in a single pass, the dequantization controller 83 outputs to the selector 81 a selection signal SEL instructing the selector 81 to select one-dimensional quantized DCT coefficients decoded by the variable length decoder 80. This causes the output terminal 91 to output two-dimensional dequantized DCT coefficients in the sequential image buildup mode.
Step S3: It is judged whether or not the progressive image buildup is in the first stage.
If the decision is affirmative, the control proceeds to step S2. Since the data structure in the first stage of the progressive image buildup, e.g.
having only a DC element and others being all zero, is the same as that of the sequential image buildup, step S2 is executed.
If the decision is negative, the control proceeds ~8~2 to step S4.
Step S4: The dequantlzation controller 83 outputs the selectlon slgnal SEL to the selector 81 lnstructlng lt to select the dummy slgnal "D0" generated by the dummy signal generator 88. The selector 81 ln turn outputs the dummy slgnal "D0" to the quantlzed coefflclent receptor 84. Step S5:
Then, the dequantlzatlon controller 83 outputs the selectlon slgnal SEL to the selector 81 lnstructlng lt to select outputs from the varlable length decoder 80. Thls change ln selectlon causes the selector 81 to output signals having the data structure shown ln Flgure 13C.
A repetition of the above steps completes the controls for processlng a whole lmage ln the sequentlal lmage bulldup mode or each stage ln the progresslve bulldup mode untll a detailed whole lmage ls obtalned.
Regardless of the stage of the progresslve lmage bulldup, the data structure of the slgnals, outputted from the selector 81 under control to the dequantlzatlon controller 83, comprlses the dlfference from the precedlng block or lts dummy slgnal "D0" at the head end, runs RM and indlces IN. When an image is reconstructed from the data structure, conventionally, the klnds of the data structures need to be declded flrst and dequantlzatlon needs to be performed accordlng to the lmage buildup modes. However, slnce this lnventlon conflgures the same clrcult to dequantize decoded DCT coefficients, it elimlnates the necesslty for such a decision.
In a progresslve image buildup, image data are reconstructed ln stage unlts from DCT coefflclents outputted from the multlpller 86 reconstructed also ln stage unlts.
After the flrst stage, the newly reconstructed lmage data are added to the hltherto reconstructed lmage data to obtaln lmage data more closely representlng the orlglnal lmage.
As descrlbed above, thls lnventlon enables a slngle common clrcult to adaptlvely dequantlze decoded one-dlmenslonal quantlzed DCT coefflclents havlng a data structure correspondlng to elther the sequentlal lmage bulldup mode or the progresslve lmage bulldup mode, thereby enabllng the clrcult to be downsized.
Flgure 16 ls a block dlagram of an lmage data reconstructlng clrcult for reconstructlng lmage data from the decoded dequantlzed DCT coefflclents outputted from the dequantlzer shown ln Flgure 14. A system controller 95 outputs to the dequantlzatlon controller 83 the speclflcatlon slgnal STA speclfylng the sequential lmage bulldup or the progresslve lmage bulldup, and controls an lmage storlng unlt 96 and an adder 97.
The two-dlmenslonal lnverse DCT unlt 44 performs lnverse DCTs on decoded dequantlzed DCT coefflclents supplled through the output termlnal 91 lnto reconstructed lmage data, whlch have two klnds. Image data DA will be used to refer to one klnd of data for the sequentlal image bulldup and the progresslve lmage buildup ln the flrst stage. Image data DB
wlll be used to refer to the other klnd of data for the progresslve lmage bulldup ln the second and later stages.
The system controller 95 constantly keeps track of e 28151 43 whlch klnd of lmage data ls outputted from the two-dlmensional lnverse DCT unlt 44.
In the case of lmage data DA, the system controller 95 halts the operatlons of the adder 97, and causes the lmage storlng unlt 96 to recelve and store outputs from the two-dlmenslonal lnverse DCT unlt 44 "as ls" at the storage addresses the system controller 95 speclfles.
In the case of lmage data DB, the system controller 95 actlvates the operatlons of the adder 97, so that the newly lnputted lmage data DB ls added to the correspondlng lmage data hltherto accumulated ln the lmage storlng unit 96, then stores the result J
in the same image storing unit 96.
The adder 97 adds image data received after stage 1 in a progressive image buildup to the image data hitherto accumulated in the image storing unit 96 for fine-tuning the reconstructed image.
Figure 16 shows a configuration in which the system controller 95 halts the operations of the adder 97 and causes the image storing unit 96 to store the outputs from the two-dimensional inverse DCT unit 44 "as is". However, an embodiment of this invention can adopt a configuration such that the system controller 95 clears the image data stored in the image storing unit 96.
Although the image data reconstructing method and apparatus described in the above embodiments assume dummy signals or decoded data are selected, it is possible configure them to select dummy signals or coded data before being decoded.
Although the image data reconstructing method and apparatus described in the above embodiments assume image processing through two-dimensional discrete cosine transforms, this invention is also applicable to image processing through other methods of orthogonal transforms.
Claims (20)
1. An image data reconstructing apparatus, for reconstructing an image from coded data obtained by coding quantized coefficients, which are obtained by quantizing transform coefficients obtained by orthogonally transforming gradient values of plural picture elements in respective blocks each comprising a plurality of picture elements divided from an original image, for progressively building up fine images from rough images in the progressive image buildup mode; said image data reconstructing apparatus comprising:
a dummy data generator for generating dummy data;
a selector for selecting either said dummy data or said coded data; and a controller for having said selector select only said coded data in a first stage, and said dummy data and then said coded data in second and later stages.
a dummy data generator for generating dummy data;
a selector for selecting either said dummy data or said coded data; and a controller for having said selector select only said coded data in a first stage, and said dummy data and then said coded data in second and later stages.
2. The image data reconstructing apparatus according to claim 1, wherein:
said coded data for said second and later stages respectively comprise a run indicating the number of zero value AC element coefficients and an index indicating the value of the non-zero value AC element coefficient; and said coded data for the first stage comprise the difference of a DC element coefficient of the current block from that of the preceding block, as well as a run indicating the number of zero value AC element coefficients and an index indicating the value of the non-zero value AC element coefficient.
said coded data for said second and later stages respectively comprise a run indicating the number of zero value AC element coefficients and an index indicating the value of the non-zero value AC element coefficient; and said coded data for the first stage comprise the difference of a DC element coefficient of the current block from that of the preceding block, as well as a run indicating the number of zero value AC element coefficients and an index indicating the value of the non-zero value AC element coefficient.
3. The image data reconstructing apparatus according to claim 1, wherein:
said dummy data indicates a DC element coefficient is zero.
said dummy data indicates a DC element coefficient is zero.
4. An image data reconstructing method for reconstructing an image from coded data obtained by coding quantized coefficients, which are obtained by quantizing transform coefficients obtained by orthogonally transforming gradient values of plural picture elements in respective blocks each comprising a plurality of picture elements divided from an original image, in which either a sequential image buildup mode for sequentially building up images or a progressive image buildup mode for progressively building up fine images from rough images is selected;
said image data reconstructing method comprising:
a first step of decoding coded data into one-dimensional quantized transform coefficients;
a second step of creating one-dimensional quantized transform coefficients having a common data structure for use by both the sequential image buildup mode and the progressive image buildup mode;
a third step of transforming said one-dimensional quantized transform coefficients created in said second step into two-dimensional quantized transform coefficients;
a fourth step of storing thresholds in preparation for dequantizing quantized transform coefficients into dequantized transform coefficients;
and a fifth step of dequantizing said two-dimensional quantized transform coefficients obtained in said third step by said thresholds stored in said fourth step; wherein:
said thresholds stored in said fourth step sequentially dequantize said quantized transform coefficients created in said second and third steps regardless of the image buildup mode in said fifth step.
said image data reconstructing method comprising:
a first step of decoding coded data into one-dimensional quantized transform coefficients;
a second step of creating one-dimensional quantized transform coefficients having a common data structure for use by both the sequential image buildup mode and the progressive image buildup mode;
a third step of transforming said one-dimensional quantized transform coefficients created in said second step into two-dimensional quantized transform coefficients;
a fourth step of storing thresholds in preparation for dequantizing quantized transform coefficients into dequantized transform coefficients;
and a fifth step of dequantizing said two-dimensional quantized transform coefficients obtained in said third step by said thresholds stored in said fourth step; wherein:
said thresholds stored in said fourth step sequentially dequantize said quantized transform coefficients created in said second and third steps regardless of the image buildup mode in said fifth step.
5. The image data reconstructing method according to claim 4, wherein:
said second step also adds dummy signals to head ends of said one-dimensional quantized transform coefficients for said second and later stages decoded in said first step during a progressive image buildup.
said second step also adds dummy signals to head ends of said one-dimensional quantized transform coefficients for said second and later stages decoded in said first step during a progressive image buildup.
6. The image data reconstructing method according to claim 5, wherein: said dummy signals are transform coefficients having zero values.
7. An image data reconstructing apparatus, for reconstructing an image from coded data obtained by coding quantized coefficients, which are obtained by quantizing transform coefficients obtained by orthogonally transforming gradient values of plural picture elements in respective blocks each comprising a plurality of picture elements divided from an original image, in which either a sequential image buildup mode for sequentially building up images or a progressive image buildup mode for progressively building up fine images from rough images is selected:
said image data reconstructing apparatus comprising:
a dequantization controller as an image buildup mode specifier for selecting either the sequential image buildup mode for sequentially reconstructing an image or the progressive image buildup mode for progressively reconstructing a fine image from a rough image;
a variable length decoder for decoding coded data into one-dimensional quantized transform coefficients;
a quantized coefficient creator for creating from one-dimensional quantized transform coefficients decoded by said variable length decoder one-dimensional quantized transform coefficients having a common data structure for both the sequential image buildup mode and the progressive image buildup mode;
a quantized coefficient receptor for transforming one-dimensional quantized transform coefficients into two-dimensional quantized transform coefficients and for storing the latter;
a quantization threshold storing unit for storing thresholds for dequantizing quantized transform coefficients; and a multiplier as a dequantization calculator for multiplying for dequantizations two-dimensional quantized orthogonal transform coefficients stored in said quantized coefficient receptor by said thresholds stored in said quantization threshold storing unit:
wherein:
one-dimensional quantized transform coefficients having a data structure common to both the sequential image buildup mode and the progressive image buildup mode are created for a specification of either one.
said image data reconstructing apparatus comprising:
a dequantization controller as an image buildup mode specifier for selecting either the sequential image buildup mode for sequentially reconstructing an image or the progressive image buildup mode for progressively reconstructing a fine image from a rough image;
a variable length decoder for decoding coded data into one-dimensional quantized transform coefficients;
a quantized coefficient creator for creating from one-dimensional quantized transform coefficients decoded by said variable length decoder one-dimensional quantized transform coefficients having a common data structure for both the sequential image buildup mode and the progressive image buildup mode;
a quantized coefficient receptor for transforming one-dimensional quantized transform coefficients into two-dimensional quantized transform coefficients and for storing the latter;
a quantization threshold storing unit for storing thresholds for dequantizing quantized transform coefficients; and a multiplier as a dequantization calculator for multiplying for dequantizations two-dimensional quantized orthogonal transform coefficients stored in said quantized coefficient receptor by said thresholds stored in said quantization threshold storing unit:
wherein:
one-dimensional quantized transform coefficients having a data structure common to both the sequential image buildup mode and the progressive image buildup mode are created for a specification of either one.
8. The image data reconstructing apparatus according to claim 7, wherein:
said quantized coefficient creator has a dummy signal generator for generating dummy signals to be attached to the head ends of data structures of said one-dimensional quantized transform coefficients decoded by said variable length decoder during a progressive image buildup.
said quantized coefficient creator has a dummy signal generator for generating dummy signals to be attached to the head ends of data structures of said one-dimensional quantized transform coefficients decoded by said variable length decoder during a progressive image buildup.
9. The image data reconstructing apparatus according to claim 8, wherein:
dummy signals generated by said dummy signal generator are transform coefficients having zero values.
dummy signals generated by said dummy signal generator are transform coefficients having zero values.
10. The image data reconstructing apparatus according to claim 7, wherein:
said quantized coefficient creator has a selector, which may be a multiplexer, for selecting the one-dimensional quantized transform coefficients from said variable length decoder and dummy signals from said dummy signal generator; and when said dequantization controller specifies a progressive image buildup, said selector attaches dummy signals to head ends of data structures of one-dimensional quantized transform coefficients outputted from said variable length decoder in the second and later stages.
said quantized coefficient creator has a selector, which may be a multiplexer, for selecting the one-dimensional quantized transform coefficients from said variable length decoder and dummy signals from said dummy signal generator; and when said dequantization controller specifies a progressive image buildup, said selector attaches dummy signals to head ends of data structures of one-dimensional quantized transform coefficients outputted from said variable length decoder in the second and later stages.
11. The image data reconstructing apparatus according to claim 1, wherein:
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
12. The image data reconstructing apparatus according to claim 2, wherein:
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
13. The image data reconstructing apparatus according to claim 3, wherein:
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
14. The image data reconstructing method according to claim 4, wherein:
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
15. The image data reconstructing method according to claim 5, wherein:
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
16. The image data reconstructing method according to claim 6, wherein:
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
17. The image data reconstructing apparatus according to claim 7, wherein:
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
18. The image data reconstructing apparatus according to claim 8, wherein:
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
19. The image data reconstructing apparatus according to claim 9, wherein:
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
20. The image data reconstructing apparatus according to claim 10, wherein:
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
said orthogonal transform is a two-dimensional discrete cosine transform; and said transform coefficients are DCT coefficients.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP02-213379 | 1990-08-10 | ||
| JP21337990A JP2839943B2 (en) | 1990-08-10 | 1990-08-10 | Image data restoration method and apparatus |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CA2048882A1 CA2048882A1 (en) | 1992-02-11 |
| CA2048882C true CA2048882C (en) | 1996-11-12 |
Family
ID=16638223
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA 2048882 Expired - Fee Related CA2048882C (en) | 1990-08-10 | 1991-08-09 | Image data processing method and apparatus |
Country Status (6)
| Country | Link |
|---|---|
| US (1) | US5369502A (en) |
| EP (2) | EP0470851B1 (en) |
| JP (1) | JP2839943B2 (en) |
| KR (1) | KR940009117B1 (en) |
| CA (1) | CA2048882C (en) |
| DE (2) | DE69132552T2 (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR0121328B1 (en) * | 1991-12-13 | 1997-11-17 | 사또오 후미오 | Digital video signal recording/reproducing apparatus |
| JPH06141185A (en) * | 1992-10-26 | 1994-05-20 | Nec Corp | Picture thinning device |
| JP3007235B2 (en) * | 1992-11-10 | 2000-02-07 | 富士写真フイルム株式会社 | Variable length code decompression device and compression / decompression device |
| JPH08163594A (en) * | 1994-12-12 | 1996-06-21 | Sony Corp | Moving image decoding method and moving image decoder |
| US6339614B1 (en) * | 1996-12-20 | 2002-01-15 | Intel Corporation | Method and apparatus for quantizing and run length encoding transform coefficients in a video coder |
| JP4527322B2 (en) * | 2001-07-25 | 2010-08-18 | 日本電気株式会社 | Image search device, image search method, and image search program |
| US6714203B1 (en) * | 2002-03-19 | 2004-03-30 | Aechelon Technology, Inc. | Data aware clustered architecture for an image generator |
| JP4486847B2 (en) * | 2003-06-16 | 2010-06-23 | オセ−テクノロジーズ・ベー・ヴエー | Method and apparatus for creating halftone images from compressed images |
| JP4725127B2 (en) * | 2005-02-16 | 2011-07-13 | ソニー株式会社 | Decoding apparatus and method, recording medium, and program |
| JP6210368B2 (en) * | 2012-09-18 | 2017-10-11 | サン パテント トラスト | Image decoding method and image decoding apparatus |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2962722B2 (en) * | 1988-02-22 | 1999-10-12 | キヤノン株式会社 | Decryption device |
| JPH02161874A (en) * | 1988-05-16 | 1990-06-21 | Nec Corp | Picture signal coding system and device therefor |
| JPH0683442B2 (en) * | 1988-05-16 | 1994-10-19 | 日本電気株式会社 | Image signal coding method and apparatus |
| JP2826321B2 (en) * | 1988-07-23 | 1998-11-18 | 日本電気株式会社 | Orthogonal transform coding device |
| JPH0810935B2 (en) * | 1988-12-09 | 1996-01-31 | 富士写真フイルム株式会社 | Image signal compression encoding device and decompression reproducing device |
| JP2571274B2 (en) * | 1989-02-14 | 1997-01-16 | 富士通株式会社 | Code data storage / readout method |
| US4954892A (en) * | 1989-02-14 | 1990-09-04 | Mitsubishi Denki Kabushiki Kaisha | Buffer controlled picture signal encoding and decoding system |
| JPH02274028A (en) * | 1989-04-17 | 1990-11-08 | Nec Corp | Image transmission method |
| US5121216A (en) * | 1989-07-19 | 1992-06-09 | Bell Communications Research | Adaptive transform coding of still images |
| JPH0832039B2 (en) * | 1989-08-19 | 1996-03-27 | 日本ビクター株式会社 | Variable length coding method and apparatus thereof |
| US5150432A (en) * | 1990-03-26 | 1992-09-22 | Kabushiki Kaisha Toshiba | Apparatus for encoding/decoding video signals to improve quality of a specific region |
| KR930011694B1 (en) * | 1991-03-27 | 1993-12-18 | 삼성전자 주식회사 | Compressing method and apparatus for digital video data |
-
1990
- 1990-08-10 JP JP21337990A patent/JP2839943B2/en not_active Expired - Fee Related
-
1991
- 1991-08-09 DE DE1991632552 patent/DE69132552T2/en not_active Expired - Fee Related
- 1991-08-09 DE DE1991628054 patent/DE69128054T2/en not_active Expired - Fee Related
- 1991-08-09 CA CA 2048882 patent/CA2048882C/en not_active Expired - Fee Related
- 1991-08-09 EP EP19910307336 patent/EP0470851B1/en not_active Expired - Lifetime
- 1991-08-09 EP EP19970103965 patent/EP0786903B1/en not_active Expired - Lifetime
- 1991-08-10 KR KR1019910013815A patent/KR940009117B1/en not_active IP Right Cessation
-
1994
- 1994-03-21 US US08/210,748 patent/US5369502A/en not_active Expired - Lifetime
Also Published As
| Publication number | Publication date |
|---|---|
| EP0786903A2 (en) | 1997-07-30 |
| EP0470851B1 (en) | 1997-10-29 |
| EP0786903A3 (en) | 1997-10-29 |
| DE69128054T2 (en) | 1998-02-26 |
| JP2839943B2 (en) | 1998-12-24 |
| JPH0495469A (en) | 1992-03-27 |
| EP0470851A2 (en) | 1992-02-12 |
| DE69132552T2 (en) | 2001-06-21 |
| EP0470851A3 (en) | 1993-06-09 |
| US5369502A (en) | 1994-11-29 |
| DE69128054D1 (en) | 1997-12-04 |
| DE69132552D1 (en) | 2001-04-12 |
| CA2048882A1 (en) | 1992-02-11 |
| EP0786903B1 (en) | 2001-03-07 |
| KR920005014A (en) | 1992-03-28 |
| KR940009117B1 (en) | 1994-09-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP0789494B1 (en) | Coding apparatus | |
| US6198848B1 (en) | Method and apparatus for compressing and storing data indicative of a full-color image | |
| Minami et al. | An optimization approach for removing blocking effects in transform coding | |
| US5737451A (en) | Method and apparatus for suppressing blocking artifacts in block-transform coded images | |
| US5719961A (en) | Adaptive technique for encoder and decoder signal transformation | |
| EP0360502B1 (en) | Efficient image signal coding system | |
| CA2048882C (en) | Image data processing method and apparatus | |
| US5748795A (en) | Image decoder using adjustable filtering | |
| US7187805B1 (en) | Maximum likelihood estimation of JPEG quantization values | |
| JP2003198851A (en) | Signaling adaptive-quantization matrices in jpeg using end-of-block codes | |
| JPH06292019A (en) | Picture data compressor and picture code compressor | |
| EP0719052A2 (en) | Image coding method and apparatus with code amount estimation | |
| CA2046544C (en) | Image data encoding system | |
| JPH0575875A (en) | Picture encoding device | |
| JPH10304362A (en) | Image signal processing unit and its method | |
| JPH06164941A (en) | Picture data coding method and device | |
| JP3394619B2 (en) | Image compression system | |
| JPH10126778A (en) | Image compression method and its equipment | |
| Helsingius et al. | Image compression using ac prediction | |
| JP2000165877A (en) | Image coder and image decoder | |
| JPH046953A (en) | Picture coding system | |
| JPH1028268A (en) | Method for reducing image-encoded data amount | |
| Wong et al. | Adaptive discrete cosine transformation of pictures using an energy distribution logarithmic model | |
| JPH09205552A (en) | Image coding method and device therefor | |
| KR0183150B1 (en) | Fractal encoding apparatus and method of variable block size without repetitive application |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| EEER | Examination request | ||
| MKLA | Lapsed |