DATA PROCESSING METHODS AND APPARATUS IN DIGITAL DISPLAY SYSTEMS
TECHNICAL FIELD OF THE INVENTION [0001] The present invention is related generally to the art of digital display systems using spatial light modulators such as micromirror arrays or ferroelectric LCD arrays, and more particularly, to a method and an apparatus for converting a stream of image data from a pixel-by-pixel format into bitplane-by-bitplane format
BACKGROUND OF THE INVENTION [0002] In current digital display systems using micromirror arrays or other similar spatial light modulators such as ferroelectric LCDs, each pixel of the array is individually addressable and switchable between an ON state and an OFF state In the ON state, the micromirror reflects incident light so as to generate a "bright" pixel on a display target In the OFF state, the micromirror reflects the incident light so as to generate a "dark" pixel on the display target Grayscale images can be created by turning the micromirror on and off at a rate faster than the human eye can perceive, such that the pixel appears to have an intermediate intensity proportional to the fraction of the time when the micromirror is on This method is generally referred to as pulse-width-modulation (PWM) Full-color images may be created by using the PWM method on separate SLMs for each primary color, or by a single SLM using a field-sequential color method
[0003] For addressing and turning the micromirror on or off, each micromirror may be associated with a memory cell circuit that stores a bit of data that determines the ON or OFF state of the micromirror In order to achieve various levels of perceived light intensity by human eyes using PWM, each pixel of a grayscale image is represented by a plurality of data bits Each data bit is assigned a significance Each time the micromirror is addressed, the value of the data bit determines whether the addressed micromirror is on or off The bit significance determines the duration of the micromirror's on or off period The bits of the same significance from all pixels of the image are called a bitplane If the elapsed time the micromirrors are left in the state corresponding to each bitplane is proportional to the relative bitplane significance, the micromirrors produce the desired grayscale image
[0004] In practice, the memory cells associated with the micromirror array are loaded with a bitplane at each designated addressing time During a frame penod, a number of bitplanes are loaded into the memory cells for producing the grayscale image, wherein the number of bitplanes equals the predetermined number of data bits representing the image pixel
[0005] The bitplane-by-bitplane formatted image data (hereafter, bitplane data), however, are not immediately available from peripheral image sources, such as a video camera, DVD/VCD player, TV/HDTV tuner, or PC video card, because the outputs (thus the input for the memory cells) of the image sources are usually either pixel-by-pixel formatted data (hereafter, pixel data), in which all bits of a single pixel are presented simultaneously, or standard analog signals that are digitized and transformed into pixel data Pixel data is typically provided as a set of parallel signals, each of which carries a bit of different significance All bits of a
particular pixel are presented simultaneously across the set of signals Successive pixels in the image are presented sequentially in time, typically synchronized with a pixel clock which is either provided by the image source or derived from other timing signals provided by the image source (such as horizontal- and vertical-sync signals) The pixel-by-pixel data format for the stream of video data is natural for non-PWM display technologies such as CRTs or analog LCDs, and has become the standard format for video data due to the historical dominance of these technologies In order for PWM-based digital displays to interface with pixel-by-pixel formatted image sources, it is necessary to reformat the incoming video data (e g the pixel data) such that the bitplanes of the image can be stored and retrieved efficiently
[0006] Therefore, methods and apparatus are desired for transforming a stream of pixel data into bitplane data
SUMMARY OF THE INVENTION [0007] In view of the foregoing, the present invention provides a method and apparatus of converting a stream of pixel data in space and time into a stream of bitplane data The apparatus of the present invention receives the pixel data in a "real-time' fashion, and dynamically performs predefined permutations so as to accomplish the predefined transpose operation In another embodiment of the invention, the pixel data are stored in a storage medium, and the apparatus of the present invention retrieves the pixel data and performs the predefined permutation to accomplish the predefined transpose operation The methods and apparatus disclosed herein are especially useful for processing a high-speed stream of digital data in a flow-through manner and suitable for implementation in a hardware video pipeline The control signal fanout and gate count of this invention are reduced compared to currently available similar techniques for converting pixel data into bitplane data
[0008] In an embodiment of the invention, a method is disclosed The method is used in a display system that comprises an array of micromirrors, each micromirror being associated with a memory cell of a memory cell array, to produce images The method comprises loading a pixel data matrix of the image, the pixel data matrix following a pixel data format, wherein matrix elements in one column of the matrix represent one pixel of the image, delivering the rows of the matnx in parallel into a data converter, transposing, by the data converter, the pixel data matrix into a bitplane matrix following a bitplane format wherein matrix elements in one row of the matrix represent one pixel of the image, and sending the bitplane matrix into the memory cell array for actuating the micromirrors such that the image is produced by the micromirrors
[0009] In another embodiment of the invention, a method is disclosed The method is used in a display system that compnses an array of micromirrors, each micromirror being associated with a memory cell of a memory cell array, to produce images The method comprises deliveπng an input pixel data matrix having m columns and n rows of the image to a data converter such that the rows of the pixel data matnx are delivered in parallel into the data converter, and the columns are delivered to the data converter sequentially at a series of
time-units, delaying the data elements of the input matrix such that the pixel data at column / and rowy (where 1<=ι<=m and 1<=j<=n) is delayed y-1 time-units relative to the data at column / and row 1, shifting the delayed data elements at each time-unit of the sequence of time-units according to a shifting rule, wherein the shifting rule states that a) the data element of row; at the /f"1 time-unit of the time-unit sequence is shifted to row ((n+j-k- 1) mod k)+1 at the same time-unit, wherein k runs from 1 to m+n time-units, and delaying the shifted data elements according to the sequence of time-units such that a data element of row j is delayed n-y time-units relative to the data element of row n
[0010] In yet another embodiment of the invention, a method is disclosed The method is used in a display system that comprises an array of micromirrors, each micromirror being associated with a memory cell of a memory cell array, to produce images And the method comprises delivering a pixel data matπx of the image to a data converter such that the rows of the pixel data matrix are delivered in parallel into the data converter, wherein the pixel data matπx follows a pixel data format, transforming the pixel data matrix into a block matrix having 2x2 first order blocks, each first order block having 2x2 second order blocks, each second order block having 2x2 third order blocks, each k order block having 2x2 (/c+1)"1 order blocks, and the (n-1)"1 order block having 2x2 pixel data elements, transposing the pixel data matπx based on the (π-1)"1 order blocks, each of which has 2x2 pixel data elements, transposing the pixel data matrix based on the k>h order blocks after consecutive transposes of the pixel data matrix based on the (n-1)"1 order block through the (/c+1)"1 order blocks, and transposing the pixel data matπx based on the first order blocks
[0011] In yet another embodiment of the invention, an apparatus is provided The apparatus is used in a display system that comprises an array of micromirrors, each micromirror being associated with a memory cell of a memory cell array to produce images The apparatus comprises a plurality of n input lines that are associated with a sequence of time-units, each input line being designated for receiving a row of data elements of a pixel data matπx, a delay unit connected to the plurality of input lines, wherein the delay unit delays each line of received data such that a) the data element at input line j is delayed one time- unit relative to the data element at input line j-1 , b) a shifter connected to and receiving output data from the delay unit, wherein the shifter shifts the delayed data output from the first delay unit based on the sequence of time-units and according to a shifting rule, wherein the shifting rule states that the data element of line; at the kfh time-unit is shifted to line ((n+j-k-1) mod k)+1 at the same time-unit, wherein k runs from 1 to m+n time-units
[0012] In yet another embodiment of the invention, an apparatus is provided The apparatus is used in a display system that comprises an array of micromirrors, each micromirror being associated a memory cell of a memory cell array to produce images The apparatus comprises a plurality of input lines that are associated with a sequence of time-units, each input line being designated for receiving a row of data elements of a pixel data matπx having m columns and n rows, a multiplicity of sets of delay units, a) wherein a delay unit of the first set of delay units delays a data element one time-unit, and the delay units of the first set are connected to every two input lines, and b) wherein a delay unit of the s,Λ set of delay unit delays a data 2s 1 time-units, and
the delay units of the sth set are connected to every 2s 1 input lines, a plurality of sets of switches, a )whereιn a switch of the first set of switches exchanges data elements between input lines 2s-1 and 2s with s running from 1 to nil, and b) wherein a switch of the s(Λ set of switches exchanges or passes through data elements between w and w +2s 1, where 1<=w<=2s 1, and wherein each switch of the s"1 set of switches are located between and connected to two delay units of the s'Λ set of delay units
BRIEF DESCRIPTION OF DRAWINGS [0013] While the appended claims set forth the features of the present invention with particularity, the invention, together with its objects and advantages, may be best understood from the following detailed description taken in conjunction with the accompanying drawings of which
[0014] FIG 1 illustrate an exemplary display system using a spatial light modulator having an array of micromirrors,
[0015] FIG 2 is a diagram schematically illustrating a cross-sectional view of a portion of a row of the micromirror array and a controller connected to the micromirror array for controlling the states of the micromirrors of the array,
[0016] FIG 3 illustrates an exemplary stream of pixel data and an exemplary stream of bitplane data, [0017] FIG 4 illustrates an exemplary converter that converts a pixel data into a bitplane data according to an embodiment of the invention,
[0018] FIG 5 illustrates an exemplary method of transforming a square matπx into another square matrix having sub-blocks with each sub-block being a 2x2 matπx,
[0019] FIG 6 illustrates another exemplary converter that converts a pixel data matrix of FIG 6 into a bitplane data matrix of FIG 6 according to another embodiment of the invention,
[0020] FIG 7 illustrates another exemplary converter that converts a pixel data matrix of FIG 6 into a bitplane data matrix of FIG 6 according to another embodiment of the invention,
[0021] FIG 8 illustrates yet another exemplary converter that converts a pixel data into a bitplane data according to yet another embodiment of the invention,
[0022] FIG 9 illustrates yet another exemplary converter that converts a pixel data into a bitplane data according to yet another embodiment of the invention,
[0023] FIG 10 illustrates matrices at different stages of a conversion executed by the converter of FIG 9, and [0024] FIG 11 illustrates an exemplary shift-register used by the converter of FIG 9
DETAILED DESCRIPTION OF THE EMBODIMENTS [0025] Embodiments of the present invention can be implemented in a variety of ways and display systems In the following embodiments of the present invention will be discussed in a display system that employs a
micromirror array and a pulse-width-modulation technique, wherein individual micromirrors of the micromirror array are controlled by memory cells of a memory cell array It will be understood by those skilled in the art that the embodiments of the present invention are applicable to any grayscale or color pulse-width-modulation methods or apparatus, such as those described in US patent 6,388,661, and US patent application serial number 10/340,162, filed January 10, 2003, both to Richards, the subject matter of each being incorporated herein by reference Each memory cell of the memory cell array can be a standard 1T1C (one transistor and one capacitor) circuit Alternatively, each memory cell can be a "charge-pump-memory cell" as set forth in US patent application 10/340,162 filed January 10, 2003 to Richards, the subject matter being incorporated herein by reference A charge-pump-memory-cell compπses a transistor having a source, a gate, and a drain, a storage capacitor having a first plate and a second plate, and wherein the source of said transistor is connected to a bitline, the gate of said transistor is connected to a wordline, and wherein the drain of the transistor is connected to the first plate of said storage capacitor forming a storage node, and wherein the second plate of said storage capacitor is connected to a pump signal It will be apparent to one of ordinary skills in the art that the following discussion applies generally to other types of memory cells, such as DRAM, SRAM or latch The word nes for each row of the memory array can be of any suitable number equal to or larger than one, such as a memory cell array having multiple wordlmes as set forth in US patent application "A Method and Apparatus for Selectively Updating Memory Cell Arrays" filed April 2, 2003 to Richards, the subject matter being incorporated herein by reference For clarity and demonstration purposes only, the embodiments of the present invention will be illustrated using binary-weighted PWM waveforms It is clear that other PWM waveforms (e g other bit-depths and/or non binary weightings) may also be applied Furthermore, although not limited thereto, the present invention is particularly useful for operating micromirrors such as those described in U S patent No 5,835,256, the contents of which are hereby incorporated by reference
[0026] Turning to the drawings, FIG 1 illustrates a simplified display system using a spatial light modulator having a micromirror array, in which embodiments of the present invention can be implemented In its very basic configuration, display system 100 compπses light source 102, optical devices (e g light pipe 106, condensing lens 108 and projection lens 116), display target 118, spatial light modulator 110 that further comprises an array of micromirrors (e g micromirrors 112 and 114), and data processing unit 123 that further comprises data converter 120 Color filter 104 may be provided for creating color images
[0027] Light source 102 (e g an arc lamp) emits light through color filter 104, light integrator/pipe 106 and condensing lens 108 and onto spatial light modulator 110 Each micromirror (e g micromirror 112 or 114) of spatial light modulator 110 is associated with a pixel of an image or a video frame and is can be actuated by a controller (e g as disclosed in US patent 6,388,661 issued May 14, 2002 incorporated herein by reference) so as to reflect light from the light source either into (the micromirror at the ON state) or away from (the micromirror at the OFF state) projection optics 116, resulting in an image or a video frame on display target 118 (screen, a viewer's eyes, a photosensitive material, etc )
[0028] A micromirror typically comprises a movable mirror plate that reflects light and a memory cell disposed proximate to the mirror plate, which is better illustrated in FIG 2 Referring to FIG 2, a cross-sectional view of a portion of a row (or a column) of the micromirror array of spatial light modulator 110 in FIG 1 is illustrated therein Each mirror plate is movable and associated with an electrode and memory cell For example, mirror plate 130 is associated with memory cell 134 and an electrode that is connected to a voltage node of the memory cell In other alternative implementations, each memory cell circuit can be associated with a plurality of mirror plates Specifically, each memory cell circuit is connected to a plurality of pixels (e g mirror plates) of a spatial light modulator for controlling the state of those pixels of the spatial light modulator An electrostatic field can be established between the mirror plate and the electrode In response to the electrostatic field, the mirror plate is rotated to the ON state or the OFF state The data bit stored in the memory cell (the voltage node of the memory cell) determines the electrostatic field, thus determines whether the mirror plate is on or off
[0029] In practice, rows or partial rows of memory cells are individually addressable by a plurality of wordlmes, and the data bits stored in the memory cells are written through a plurality of bit lines, each of which is connected to a column of memory cells of the memory cell array The addressing and wπtmg the memory cells are controlled by controller 124 (e g the controller as disclosed in US patent 6,388,661 issued May 14, 2002, the subject matter being incorporated herein by reference) The controller communicates with data processing unit 123, which prepares data, such as bitplane data, for the micromirrors
[0030] In this figure, the memory cells are illustrated as standard 1T1C memory cells and only one wordline is provided for the row of memory cells It should be understood than this is not an absolute requirement Instead, other memory cells, such as a charge-pump-memory cell, DRAM or SRAM could also be used Moreover, the memory cells of each row of the memory cell array could be provided with more than one wordline for addressing the memory cells In particular, two wordlmes could be provided for each row of memory cells of the memory cell array as set forth in US patent application 10/340,162 filed January 10, 2003 to Richards, the subject matter being incorporated herein by reference
[0031] Turning back to FIG 1, the spatial light modulator having the memory cell array is connected to data processing unit 123 that further comprises data converter 120 The data processing unit receives image data from peripheral image sources, such as video camera 122 and processes the received image data into pixel data as appropriate Data converter 120 converts the pixel data into bitplane data that can be loaded into the memory cells of the memory cell array For example, when image source 122 outputs standard analog image signals, the data processing unit samples the image signals and transforms the image signals into digital pixel data Then data converter 120 converts the pixel data into bitplane data Data converter 120 as illustrated in the figure is part of the data processing unit Alternatively, the data converter can be a separate device to the data processing unit In this case the data converter is connected to the data processing unit The data converter
receives data from the data processing unit and outputs bitplane data to the memory cells of the memory cell array
[0032] Referring to FIG 3, exemplary pixel data and exemplary bitplane data are illustrated therein In this figure, symbol α* represents a binary data bit in a sense that that the data bit is either "1" or "0" "1" and "0" correspond to relative voltages of the memory cell For example, "1" and "0" respectively correspond to a high voltage and a low voltage of the memory cell Alternatively, "1" and "0" respectively correspond to the low voltage and the high voltage of the memory cell The subscription / identifies the pixel of the desired image The value of a, determines the voltage of the memory cell, thus the on or off state of the micromirror The superscript k labels the bit number (or the significance) of the pixel based on the pulse-width-modulation For example, /(could be a number from 1 to n (e g n=8) when n data bits (e g 8 bits) are used to represent different grays levels (e g 256 levels for 8 bits) of the pixel using the pulse-width-modulation technique The value of k determines the time of the voltage maintained by the memory cell
[0033] The pixel data is ordered in time by the positions of pixels of the desired image In display systems without micromirrors, data bits of the same pixel are loaded at one time for producing the pixel of the image For example, at time fi, data bits in the first column [a] , a a a" ) are loaded for producing the first pixel of the desired image At another time f„ data bits in the /"" column (α ,α a' a" ) are loaded for producing the ;)Λ pixel of the desired image
[0034] In contrast, the bit plate data is primarily ordered by bits of all pixels of the desired image In display systems using micromirrors, data bits of the same significance for all pixels of the desired image are loaded at one time for actuating the mirror plates For example, at time , the first bits of all n pixels in the first column
{aj' .ai ,' an' ) are loaded into the memory cells of the spatial light modulator for actuating the mirror plates At another time t, data bits in the " column {α,' ,α2' α^ αn' ) representing the /'" bitplane of all pixels are loaded into the memory cells
[0035] By compaπng the pixel data matrix and the bitplane data matrix, it can be seen that the bitplane data matrix is a transpose matrix of the pixel data matrix By 'data matπx", it is meant that a block of data elements that are organized into a plurality of rows and columns of data elements The data elements in each row are disposed in time sequence - that is the data elements in a row are delivered at different time units The data elements in each column are delivered at the same time
[0036] The pixel data matrix and the bitplane matrix in FIG 3 are illustrated as square matrices nxn In practice, the pixel data matrix can be a rectangular matrix with unequal numbers of rows and columns As a consequence, the bitplane data matrix as a transposed matrix of the pixel data matπx is also a rectangular matrix As illustrated in the figure, the data bits of the same pixel are arranged in the same column of the pixel data matrix and read at one time This arrangement scheme, however, is not an absolute requirement Instead,
the data bits of the same pixel may be stored in the same row of the pixel data matπx and read at the same time Accordingly, the bitplane matrix as a transposed matrix of the pixel data matrix is loaded into the memory cell array row by row
[0037] In order to convert a pixel data matrix into a bitplane matrix, all rows of the pixel data matrix are delivered into the data converter in parallel, and transpose operations are performed on the loaded rows simultaneously In the following, embodiments of the invention will be discussed with reference to figure 4 through figure 11 In particular, an embodiment of the invention will be discussed with reference to a 2χ2 pixel data matrix in FIG 4 The method for transposing 2x2 pixel data matrix in FIG 4 is extended to transpose a 2"χ2" pixel matrix, which will be discussed in detail with reference to FIG 5, FIG 6 and FIG 7 and FIG 8 Another embodiment of the invention will be introduced with reference to FIG 9 through FIG 11 This method is particularly useful for transposing a pixel matπx having non-power-of-2 numbers of rows and columns It is noted that, in practice, pixel data (and also bitplane data) in a row are delivered in time sequence - that is these data are sequentially delivered at different time units And pixel data (and also bit plane data) in a column are delivered at the same time through, for example, a bit line of the display system Moreover, in the embodiment of the invention, pixel data are "flowing through" the data processing unit of the display system The data processing unit receives pixel data in a "real-time" fashion and dynamically performs the predefined transpose operation on the received pixel data Specifically, the data processing unit receives a column of pixel data at a time-unit and dynamically performs predefined permutations to these received data so as to accomplish the transpose operation Of course, the pixel data can be stored in a storage medium, such as a frame buffer The data processing unit then retrieves the stored pixel data and performs the predefined transpose operation on the retrieved pixel data It should be understood that the embodiments as discussed in the following are for demonstration and clarity purposes only It should not be interpreted in any ways as a limitation to the present invention Rather, any suitable methods and apparatus without departing from the spirit of the present invention could be used
[0038] Referring to FIG 4, data converter 120 transposes pixel data matπx having 2x2 matnx elements into bitplane data matrix The two matrices are defined as following
The data converter comprises a plurality of input lines (e g ln[1] and ln[2]) and output lines (e g Out[1] and Out[2]) In practice the number of input lines equals or is greater than the number of rows of the pixel matrix The data converter further comprises switch 136 and a plurality of delay units, such as delay units 138a and 138b
[0039] Each delay unit delays a received data element at one time-unit, which may be a multiple of a clock- cycle For example, an XGA (1024x768) video signal typically has a pixel clock of 65 MHz, or a clock period of
15 1 nanoseconds In this case, the time-unit is preferably 15 1 nanoseconds, or a multiple of 15 1 nanoseconds An exemplary delay unit is the standard flipflop circuit The switch exchanges the data elements between the lines connected to the switch at certain time units For example, when time f is odd, the switch exchanges the data elements of the delayed input lines X[1] and X[l] such that the data element on X[1] before the switch is delivered to Y[2] after the switch And the data element on X[2] before the switch is delivered to Y[1] after the switch When f is even, X[1] is passed through to Y[1] and X[l] is passed through to Y[2] An exemplary switch (e g switch 136) is illustrated in the figure As can be seen from the figure, switch 136 consists of two juxtaposed multiplexers 137a and 137b, both are connected to an activation signal Co In the embodiment of the invention, the activation signal toggles every time-unit (e g every clock cycle when the time-unit equals one clock cycle) In response to the activation signal Co, the two multiplexers exchange input data bits and outputs exchanged data bits Of course, other suitable switch circuits may also be applied [0040] In the transpose operation, the data converter is associated with a sequence of clock cycles Specifically, the input lines and the output lines are synchronized with a sequence of time-units, each of which is a multiple of the clock cycle Data elements passing through the data converter are synchronized with the sequence of time-units thereby
[0041] According to the embodiment, data elements a,
1 and a
1, in the first row and data elements a and al in the second row are respectively delivered to the input lines ln[1] and ln[2] and synchronized with the sequence of time-units At time pi, data elements a,' and
are flowing in the input line ln[1] with the data element a,' one time-unit in front of data element al Data element a and a] are flowing in the input line ln[2] with data element a one time-unit in front of data element a Data elements al and al are synchronized and so are data elements a] and
By "data elements are synchronized", it is meant that there is no time delay between the data elements with reference to a common time sequence Data elements a] and a] then passes through delay unit 138a and are delayed one time-unit thereby As a result, data element a is synchronized with data element a] — that is data elements a and a) have the same time-unit, as shown in the figure After delay unit 138a, synchronized data elements in switch input lines X[1] and X[2] are exchanged by switch 136 depending on the state of control signal Co Specifically, data element a in line X[2) is may exchanged with data element al in line X[1], while data elements a' and al are unchanged, because data elements a and al are synchronized and data elements a,' and al have no synchronized data elements Therefore, al s, data elements
and α] are flowing in line Y/ /wιth data element
one time-unit in front of data element α) Data elements α and α] are flowing in input line Y[2] with data element α) one time-unit in front of data element α , as shown in the figure After switch 136, data elements in switch output line Y[1]
pass through delay unit 138b and are delayed one time-unit Consequently, at p4, data elements a] and a are flowing in output line Out[1] with data element a,' one time-unit in front of data element α; Data elements al and al are flowing in output line Out[2] with data element al one time-unit in front of data element al However, data elements a,' and al are synchronized And data elements α and a] are synchronized The data elements are then output from out lines Out[1 ] and Out[2] as shown in the figure As a result, the pixel data matrix is transposed into the bitplane data matrix that can be loaded into the memory cells for actuating the mirror plates for producing grayscale images
[0042] It is observed that a 4χ4 matrix transpose is equivalent to swapping two 2x2 sub-blocks along one diagonal of the matrix followed by transposing each of the four 2x2 sub-blocks A data converter for transposing 4x4 matrix can be constructed by juxtaposing two above converters for transposing 2x2 sub-blocks and concatenating the juxtaposed converters to a similar circuit that swaps the 4x4 matrix in terms of the four 2x2 sub-blocks An exemplary converter for transposing 4x4 matrices is illustrated in FIG 6, which will be discussed in detail in following
[0043] Referring to FIG 5, the pixel data matrix is a 4x4 matrix The pixel data matπx is transformed into a pixel block-matrix having 2x2 sub-blocks with each sub-block being a 2x2 sub-matrix The pixel block-matrix is then transposed in terms of the sub-blocks by swapping the two 2χ2 sub-blocks along the diagonal of the matrix After transpose of the sub-blocks, each sub-block is transposed In order to accomplish this transpose scheme, data converter 120 is provided as illustrated in FIG 6
[0044] Referring to FIG 6, data converter 120 comprises at least four input lines - ln[1], ln[2], ln[3] and ln[4], and at least four output lines - Out[1], Out[2], Out[3] and Out[4] The data converter further comprises two juxtaposed transpose circuits, each being configured for transposing 2χ2 matrices Specifically, one of the two juxtaposed circuits consists of delay units 138a and 138c and switch 136a And the other transpose circuit consists of delay units 138b and 138d and switch 136b These two juxtaposed transpose circuits are concatenated with a similar blocking transposing circuit that has delay units 146a, 146b, 146c and 146d and two switches 148a and 148b The blocking transposing circuit transposes the matnx in terms of the sub-blocks [0045] Switch 148a exchanges data elements between lines ln[2] and ln[4], and switch 148b exchanges data elements between lines ln[1] and ln[3] Switch 136a exchanges data elements between lines ln[1]anύ ln[2], and switch 136b exchanges data elements between lines ln[3] and ln[4] The switches 148a and 148b can be the same as the switch 136a or switch 136b However, this is not an absolute requirement Instead, each of the switches can be different from the other switches Control signal C> (controlling the first stage of switches) toggles ON for 2 clock cycles and OFF for 2 clock cycles, as shown in the figure, while control signal Co controls the 2x2 transposers similarly to that shown in FIG 4 Ci and Co must be appropriately delayed with respect to the data and the pipeline delays of the delay stages
[0046] In accordance with an embodiment of the invention, the data converter is associated with clock cycles Specifically, the input lines and the output lines ln[1], ln[2], ln[3] and ln[4] are synchronized with a sequence of time-units, each of which may be a clock cycle or a multiple of a clock cycle In the embodiment of the invention, the time-unit equals one clock cycle Data elements pass through the input lines of the data converter are synchronized with the sequence of time-units thereby
[0047] In the transpose operation, data elements of the pixel data matrix in each row are sequentially delivered into an input line in accordance with the sequence of time units Data elements of the pixel data matrix in separate rows are delivered into different input lines in parallel At time
and α
4 ! are delivered to the first input line with the data elements sequentially spaced with one time-unit Specifically, data element
is one time-unit in front of data element α
l'
+I Similar to the data elements in the first row, data elements α] through α
4 in the second row, data elements
through α
4 3 in the third row, and data elements α through α
4 4 in the fourth row are respectively delivered to the input lines ln[2], ln[3], and ln[4]
For the data elements in the same input line, data elements are sequentially spaced with one time-unit Data elements in the same column are synchronized
[0048] The data elements of the third row and the fourth row are then delayed two time-units for each data element by delay units 146a and 146b As a result, at time i, data elements α
3 and α," are synchronized with data elements α] and α
3' Data elements a and at are synchronized with data elements a and
[0049] After being delayed, data elements in the input lines are exchanged by switches 148a and 148b Specifically, switch 148a exchanges data elements between input lines ln[2] and ln[4], and switch 148b exchanges data elements between input lines ln[1] and ln[3) Both switches perform the exchange in response to control signal Ci, which toggles ON for 2 clock cycles and OFF for 2 clock cycles, as shown in the figure As a consequence, at P2, data elements a' ,al,a
3 ,a
3 are flowing in line A[1] Data elements a ,al ,a,
4 ,at are flowing in line A[2] Data elements a
3' ,a
4' ,a
3,a
4 are flowing in line A[3] And data elements α ,α
4
,α
4 are flowing in line A[4]
[0050] Following switches 148a and 148b, data elements in the first row and the second row are delayed two time-units by delay units 146c and 146d As a result, at 3, data elements of the first row and the second row are synchronized with the corresponding data elements in the third row and the fourth row For example, data elements and α in the first two rows (the first row and the second row) and first column are synchronized with data elements α3' and α ] in the last two (the third and the fourth) rows and first column
[0051] After delay units 146c and 146d, transpose of the pixel block matrix is complete Delay units 138a, 138b, 138c and 138d, and switched 136a and 136b then perform transpose to the sub-block matrix of the transposed pixel block matπx Table 1 summarizes the timing of the transposing operation discussed above
TABLE 1
[0052] As discussed above, data converter 120 in FIG 6 is configured such that a transposing circuit (having delay units 146a through 146d and switches 148a and 148b) for transposing dats matrices in terms of sub-blocks is disposed in front of the two juxtaposed transpose circuits, each being designated for transposing 2x2 sub- blocks Accordingly, a 4x4 pixel matπx is transposed in terms of sub-blocks followed by transposes of the sub- blocks This configuration, however, is not an absolute requirement Other suitable configurations may also be applied For example, the two transpose circuits for transposing sub-blocks can be placed in front of the transpose circuit for transposing the matrix in terms of the sub-blocks Such a configuration is illustrated in FIG 7 As can be seen from FIG 7, delay units 138a, 138b, 138c and 138d, and switches 136a and 136b are disposed in front of delay units 146a through 146d and switches 148a and 148b By this arrangement, transposes of the sub-blocks are performed before the transpose of the matrix in terms of the sub-blocks [0053] From the above discussion, it can be seen that the method and the apparatus for transposing 2x2 pixel matrices can be extended to transpose 4x4 pixel matrices This method 3nd apparatus can be recursively extended to build data converters for transposing a pixel dst3 matrix having 2"χ2" dats elements, wherein n is 3n integer larger than 2
[0054] For transposing such pixel data matrices into bitplane matrices, the 2"χ2" pixel dats matrix is first transformed into a pixel block matrix having 2x2 first order sub-blocks Each first order sub-block has four 2x2 second order sub-blocks, and each second order sub-block has four 2χ2 third order blocks By iterating such transformation method, the 2"χ2" pixel data matrix is transformed into a pixel block matrix having a plurality of sub-blocks with orders Each A"1 order sub-block has 2x2 (/(+1)fΛ order sub-blocks, and the (n-1)th order sub- block is a matπx having 2x2 pixel data elements
[0055] In accordance with an embodiment of the invention, the transformed pixel data matrix is first transposed based on the (n-1)"1 order sub-blocks, each of which has 2x2 pixel data elements following by transposing the pixel data block matrix based on the {n-2)th order sub-blocks The pixel data matrix is transposed based on the order sub-blocks after consecutive transposes of the pixel data matrix based on the (n-1 )'" order sub-blocks through the (/c+1)(Λ order sub-blocks Then the pixel data block is transposed based on the first order blocks
[0056] In order to transpose the pixel block matπx into bitplane matrix, the rows of the pixel block matπx are delivered in parallel into a data converter that comprises a plurality of input lines (e g ln[1] through ln[n]), a set of delay-unit sets (e g delay unity sets 1 through n-1) and a set of switch sets (e g switch sets 1 through n-1), as illustrated in FIG 8 The combination of the 1st delay unit set 146a and the 1st switch set 148a performs transpose of the (n-1)"1 order sub-blocks The combination of the 2nd delay unit set 146b and the 2nd switch set 148b performs transpose of the (n-2)'Λ order sub-blocks that has 2x2 pixel data elements The combination of the kth delay unit set 146c and the klh switch set 148c performs transpose on the (n-k-1)"1 order sub-blocks And the combination of the 2" ' delay unit set 146d and the 2" ' switch set 148d performs transpose on the 1st order
sub-blocks It can also be seen from the figure that, different combinations of the delay unit sets and the switch sets are disposed consecutively Specifically, the combination of the delay unit set and the switch set with a lower order immediately follows the combination of the delay unit set and the switch set with one order higher For example, the combination of the 2nd order delay unit set and the 2nd order switch set is immediately behind the combination of the 1st order delay unit set and the 1st order switch set For another example, the combination of the A"1 order delay unit set and the A* order switch set is immediately behind the combination of the (A-1),Λ order delay unit set and the (k- order switch set This arrangement allows for consecutive transposes of sub- matrices with consecutive orders Moreover, this arrangement guarantees that the transpose of the A"1 order sub-block matrix is performed after all transposes on the sub-block matrices with orders from n-1 to A [0057] Each delay unit set compπses one or more delay units (e g the delay unit 138a or 138b in FIG 4) The delay unit can be a standard flipflop circuit or a shift-register In the embodiment of the invention, the total number of the delay units in each delay unit set equals 22n 1 times the order of the delay unit set For example, the A"1 delay unit set has 22k 1 delay units, each of which delays the received data elements one time-unit Therefore, the total delayed time-units by the A"1 delay unit set is 22k 1 time units Each switch set comprises at least two switches, such as switch 136 in FIG 4 According to the embodiment, each switch set is "sandwiched" by two delay unit sets For example, the 1st switch set is dispose in the middle of two serially disposed 1st delay unit sets The A*" switch set is disposed in the middle of two serially disposed A* delay unit sets [0058] In performing the transpose of the pixel block matrix based on the (n-1)'" order sub-blocks each having 2x2 pixel data elements, each (n-1),Λ order sub-block is transposed by delaying the data elements in the second row of the (n-1)'Λ order sub-block one time-unit relative to the data elements in the first row of the (n-1)ft order block and delaying the data elements in the second column in each row one time-unit relative to the data elements in the first column of the same row of the (n-1)"1 order sub-block The delay is performed by the 1st delay unit set 146a in FIG 8 The data elements of the delayed (n-1)"1 order sub-block is then switched at each time-unit by the 1st switch set 148 in FIG 8 according to a predefined switching rule The switching rule is listed in table 2
TABLE 2
In the table, R, <-> R, represents an exchange operation by which data element in row / is conditionally exchanged with data element in row at a given time-unit based on a control signal [0059] After being switched, the data elements of the first row of the (n-1 J"
1 order sub-block are then delayed one time unit by the 1
st delay unit set
[0060] In performing the transpose of the pixel data matπx based on the first order sub-blocks, the data elements of the pixel data block matπx are delayed by the (N-1)ft1 delay unit set 146d in FIG 8 according to a sequence of time-units such that a) data elements of rows 1 through nil are not delayed, b) for data elements of rows from n/2+1 through n, data elements at column / and row; is delayed one time-units relative to the data element at column /+1 and rowy, and is delayed n!2 time-units relative to the data element at the same column and the first row The delayed data elements are then switched by the (N-1)"1 switch set 148 d in FIG 8 according to the switch rule in table 1 Specifically, the switch rule states that at each time-unit, a) exchanging the data element of row 1 with the data element of row (nil +1) at the time-unit, and b) exchanging the data element of row / with the data element of row (n/2+i) The switched data elements are then delayed according to the sequence of time-units such that a) data elements of rows n/2+1 through n are not delayed, b) for data elements of rows from 1 through nil, data elements at column / and rowy is delayed one time-unit relative to the data element at column /+1 and row , and is delayed n/2 time-units relative to the data element at the same column and the first row
[0061] After consecutively performing the transposes of sub-blocks with consecutive orders starting from n-1 to 1 by the data converter of FIG 8, the pixel data matπx having 2nχ2" pixel data elements is transposed into a bitplane data matrix
[0062] Rather than arranging the delay unit sets and the switch sets in an order as illustrated in FIG 8, the delay unit sets and the switch sets can be arranged in an inverse order Specifically, the combination of the delay unit set and the switch set with a lower order immediately in front of the combination of the delay unit set and the switch set with one order higher For example, the combination of the 2nd order delay unit set and the 2nd order switch set can be immediately in front of the combination of the 1st order delay unit set and the 1st order switch set, as long as the other delay unit sets and switch sets obey the same inversed arrangement order For another example, with the inverted arrangement order, the combination of the A"1 order delay unit set and the A"1 order switch set is immediately in front of the combination of the (A-1)!Λ order delay unit set and the (A-1)"1 order switch set And the combination of the (N-ψ delay unit set and the (N- switch set is placed in the front of the data converter - that is, pixel data elements of the pixel data matrix are delivered first into the combination of the (N-ψ delay unit set and the (N- switch set
[0063] Rather than arranging the delay units sets and the switch sets in the ascending order (as shown in FIG 8) or the descending order as discussed above, the delay units and the switch sets can be arranged randomly Specifically, combinations of the delay unit sets and the switch set of the same order can be disposed randomly along the input lines For example, the combination of the m'
h order delay unit sets and the m
th order
switch set can be disposed between a combination of the ι'
h order delay unit sets and the ι'
h order switch set and a combination of the f order delay unit sets and the f order switch set, wherein / ≠ m+1 and j ≠ m±l Accordingly, the pixel data matrix is transposed by random orders [0064] In addition to a pixel data matrix having 2"χ2" pixel data elements, the method and the data converter as discussed with reference to FIG 8 can be also be applied in transposing pixel data matrices having 2
πχm pixel data elements with m being an integer not equal to 2" into a bitplane data matrix [0065] For a pixel data matπx having 2"χm pixel data elements with m being an integer smaller than 2", a number of rows of "fake" data elements can be inserted into the pixel data matrix such that the pixel data matrix after insertion is a 2
nχ2
n pixel data matrix Each row of "fake" data elements consists 2" "fake" data elements, and (l
π-m) such rows are inserted into the pixel data matrix These "fake" data rows can be attached inserted before the first row of the pixel data matrix, or appended after the last row of the pixel data matπx, or inserted between the rows of the pixel data matπx, as long as the insert positions are memoπzed [0066] After performing the transpose method discussed above, the inserted "fake" data elements are removed from the transposed pixel data matrix having "fake" data elements As a way of example, (l
n-m) rows of "fake" data elements are appended after the mf
h row of the pixel data matrix After transpose, the "fake" data elements are located at positions from the (2"-m)
th column to the (2")"' column in each row Therefore, by truncating the columns from the [l
n-m)
th column to the (2")'" column, the bitplane matrix is obtained These 'fake' data elements may be implemented by hardwiring some inputs of the transposer to 0 or 1, this may allow some of the delay elements or parts of the switch logic to be optimized away or reduced [0067] The methods and the apparatus as discussed with reference to FIG 4 through FIG 8 can be characterized using a plurality of parameters, such as the longest path delay, the total number of flipflops, the total number of shift-registers, the total number of multiplexers and the control signal fanout The longest path delay is defined as the length of the longest combinational logic path between delay elements or l/Os, in terms of 2-ιnput multiplexers The control signal fanout is defined as the total number of loads driven by any control signal to the switches/multiplexers (e g the multiplexers 137a and 137b in FIG 4) Values of these parameters are listed in table 2 when the method and the apparatus as discussed above are employed in transposing a matrix having N columns, where N is a power of 2 In certain implementation technologies, the cost (in terms of circuit area) of a multiple-cycle delay element (i e a shift register) may be significantly less than the cost of the corresponding number of individual flipflops For example, certain FPGA architectures allow implementation of a 16-element shift register in a single logic block For this reason the table also tallies the total number of shift registers (of arbitrary length) in the design which may be more representative of the area cost of the design in such technologies TABLE 3
[0068] In practice, the pixel data matrix can be a rectangular matrix having m columns and n rows where n may not be a power of 2 A method and an apparatus for transposing such pixel data matrices will be discussed in the following with reference to FIG 9 through FIG 11 Obviously, such method and apparatus are also applicable for transposing 2nχ2" pixel data matrices and 2"χm pixel data matrices [0069] Referring to FIG 9, the data converter comprises delay unit set 140 and shifter 144, which is preferably a barrel shifter Under control of a set of control signals, the barrel shifter provides on its output a circularly rotated version of its inputs, where the number of positions the data is rotated is determined by the control inputs An exemplary barrel shifter is illustrated in FIG 11 Referring to FIG 11 , the barrel shifter comprises N inputs, represented by ln[1], ln[2], through ln[N], and N outputs, represented by Out[1] through Out[N] In response to a control signal "Q", the N input data are circularly rotated with Q positions as shown in the figure, wherein Q is an integer less than N Referring back to FIG 10, for simplicity and demonstration purposes only, only four input lines and four out lines are illustrated in the barrel shifter in the figure [0070] According to the embodiment of the invention, delay unit set 140 comprises a set of delay units, such as the delay unit 138a or 138b in FIG 4 For example, the delay unit can be a flipflop circuit or a shift register The delay units are deployed along the input lines of the data converter such that A number of delay units are disposed along the kf input line in front of shifter 144, and another A number of delay units are disposed along the A"7 input line after shifter 144 For example, one delay unit is disposed along the first input line ln[1] in front of shift 144 and another delay unit is disposed along the first input line after shifter 144 For another example, three delay units are disposed along the third input line ln[3] in front of shift 144 and another three delay units are disposed along the third input line after shifter 144 [0071] For simplicity and demonstration purposes, the transposing method using the data converter in FIG 9 will be discussed with reference to transposing a rectangular matπx having eight columns and four rows as illustrated in FIG 10 [0072] In the transpose operation, the data converter is associated with clock cycles Specifically, the input lines are synchronized with a sequence of time-units, each being a multiple of a clock cycle As a result, data elements flowing through the input lines are synchronized with the sequence of time units [0073] The four rows are separately connected to the four input lines - ln[1], ln[2], ln[3] and ln[4] such that pixel data elements of separate rows are delivered into the input lines of the data converter in parallel The pixel data elements in each row are delivered sequentially into an input line such that the adjacent pixel data elements in a row have one time-unit difference in time relative to each other Specifically, data element of rowy is delayed one time-unit relative to data element +; of the same row For example, a is delayed one time-unit
relative to data element a ; Data elements of the same column are synchronized with one time-unit The data elements then pass through delay unit set 140 located in front of shifter 144 and are delayed thereby Consequently, a pixel data at column / and rowy is delayed (j-1) time-units relative to the data at column ; and the first row, and one time-unit relative to the data element at column ;+1 and rowy as shown in the matrix at position T2 The delayed data elements are then shifted by shifter 144 according to the sequence of time-units and based on a shifting rule In the embodiment of the invention, the shifting rule states that for a matrix having m columns and n rows, a) the data element of rowy at the A"1 time-unit of the time-unit sequence is shifted to row ((n+y-A-1) mod n)+1 at the same time-unit, wherein A runs from 1 to m+n time-units An exemplary shift operation is illustrated in FIG 11 As shown in FIG 11, a sequence of data bits in a register is shifted rightwards sequentially The data bit at the n"1 position is shifted to the 0th position of the register Turning back to FIG 9, the shifted data elements are delayed by delay unit set 140 located behind shift 140 Similar to the delay process in the delay unit set in front of shifter 140, the shifted data elements are shifted according to the sequence of time-units such that a data element of rowy at time-unit p is delayed j-1 time-units relative to the data element of row 1 at time-unit p After the second shift, the mxn pixel data matrix is transposed and the bitplane data matrix at position T is obtained, as shown in the figure The bitplane data of the bitplane data matrix can be loaded into the memory cells for actuating the mirror plates of the micromirror array within the spatial light modulator
[0074] The methods as discussed with reference to FIG 9 and FIG 10 can be characterized using a plurality of parameters, such as the longest path delay, the total number of flipflops, the total number of shift-registers, the total number of multiplexers and the control signal fanout Values of these parameters are listed in table 3 when the method and the apparatus as discussed above are employed in transposing a matrix having N columns TABLE 4
[0075] Other than implementing the embodiments of the present invention in data converter 120 in FIG 1, the embodiments of the present invention may also be implemented in a microprocessor-based programmable unit, and the like, using instructions, such as program modules, that are executed by a processor Generally, program modules include routines, objects, components, data structures and the like that perform particular tasks or implement particular abstract data types The term "program" includes one or more program modules When the embodiments of the present invention are implemented in such a unit, it is preferred that the unit communicates with the controller, takes corresponding actions to signals, such as actuation signals from the controller, and inverts polarity of the voltage differences
[0076] It will be appreciated by those skilled in the art that a new and useful method and apparatus for transposing pixel data matrices into bitplane data matrices for use in display systems having micromirror arrays have been described herein. In view of many possible embodiments to which the principles of this invention may be applied, however, it should be recognized that the embodiments described herein with respect to the drawing figures are meant to be illustrative only and should not be taken as limiting the scope of invention. For example, those of skill in the art will recognize that the illustrated embodiments can be modified in arrangement and detail without departing from the spirit of the invention. Therefore, the invention as described herein contemplates all such embodiments as may come within the scope of the following claims and equivalents thereof.