WO2005115006A1

WO2005115006A1 - System and methods of encoding moving pictures for mobile communication terminals

Info

Publication number: WO2005115006A1
Application number: PCT/KR2004/001203
Authority: WO
Inventors: Changho Lee
Original assignee: Multivia Co., Ltd.
Priority date: 2004-05-20
Filing date: 2004-05-20
Publication date: 2005-12-01
Also published as: CN1720745A; CN100466733C

Abstract

The present invention relates to systems and methods of encoding moving pictures for mobile communication terminals. The present invention includes: the intra-coding process that discrete cosine transforms the moving picture compression object signal, obtains quantization coefficient by quantizing the discrete cosine transformed moving picture compression object signal, Golomb-Rice codes the quantization coefficient as an entropy coding, and outputs the Golomb-Rice coded quantization coefficient; and the inter-coding process that estimates whether motions happened in a macro block unit for the moving picture compression object signal, classifies the macro block into a motion block and a non-motion block, stores the motion block and the non-motion block separately, encodes and outputs information about whether motions happened or not as texture map information, and performs an estimation of an infra mode and an inter mode for the motion block at the same time when estimating whether motions happened.

Description

SYSTEMS AND METHODS OF ENCODING MOVING PICTURES FOR MOBILE COMMUNICATION TERMINALS

TECHNICAL FIELD The present invention relates to moving picture encoding. More particularly, the present invention relates to systems and methods of encoding moving pictures for mobile communication terminals, which are capable of minimizing processes and the number of operations when encoding moving pictures and are adaptable to mobile communication terminals by embodying it in a software manner.

BACKGROUND ART Recently, technologies for processing moving pictures have been used in diverse fields. A Video On Demand (VOD) service such as a movie service on the Internet is a representative example ofthe technologies. International standards in such diverse fields for processing moving pictures are defined. As typical international standards, there are Moving Picture Experts Group (MPEG) -1 used as a compression method for a video Compact Disk (CD) storage, MPEG- 2 applied and used as a compression method for a high definition digital TV broadcasting or a Digital Video Disk (DVD) service and MPEG-4 which is a method of making a compression coder suitable for various contents and used in a moving picture compression solution in wireless environments such as an internet broadcasting or International Mobile Telecommunication-2000 (TMT-2000). In addition, there are H.261 which is developed for a video conference, has a performance similar to that of MPEG-1 and is mainly used in ISDN network, H.263 which is developed for a videophone and provides a base of MPEG-4, and H.26L (H.264) which is the most recent standardized compression method for a picture phone, a moving picture-supporting cellular phone and TV, etc. and can increase a compression performance to two times as much as MPEG-4. Meanwhile, the above-mentioned moving picture-processing technologies are applied to the various fields, and a personal computer (PC) is a field commonly contacted by the public. However, the PC has a problem of a mobility restraint. Recently, technologies of processing moving pictures in mobile communication terminals such as a Personal Digital Assistant (PDA) and a cellular phone, etc., which allows various multimedia services to be provided through an internet connection and a camera mounting and solves the mobility restraint ofthe PC, are gradually expanded. However, since the mobile communication terminal has smaller CPU performance, smaller memory space and a limited battery capacity for supplying a driving power, compared to the PC, it has many restraints when applying moving picture-processing technologies requiring many operations to the terminal. Accordingly, the moving picture- processing technologies have not been embodied in a software manner like in the PC. Instead, the moving picture-compressing technologies have been provided by a hardware chip. Recently, as performance of CPU used in the mobile communication terminal has been improved (for example, commercialization of MSM6000 from Qualcomm), there are attempts to apply the moving picture-processing technology to the mobile communication terminal in a software manner. However, since the related art moving picture-processing technology requires many operations to encode moving pictures, compared to a decoding of the moving pictures, the moving picture-processing technology cannot be actually applied to the mobile communication terminal in a software manner. Accordingly, in order to apply the moving picture-processing technology to the mobile communication terminal in a software manner, it is required to develop a moving picture compression-encoding technology capable of reducing the number of operations to meet a CPU performance of a mobile communication terminal. For example, as illustrated in Figure 1, a moving picture encoding apparatus according to the related art using a MPEG comprises an intra-coding section 20 which performs a compression only with a frame itself using a spatial correlation in a frame, and an inter-coding section 30 which performs a compression using a time relation between a current frame and a previous frame. At this time, the system encodes a moving picture using a discrete cosine transform (DCT) method. The intra-coding section 20 applies a Huffman coding method as an entropy coding when performing an encoding of MPEG series. The inter-coding section 30 applies the Huffman coding method as an entropy coding as with the intra-coding section 20, and performs an encoding using a time relation between a current frame and a previous frame for a predictive frame coding (i.e., P frame coding). At this time, in the case of an image for the previous frame, since comparison is made for an image restored from compressed image, a decoding is also performed when performing an encoding. In addition, a motion estimation unit 31 in the inter-coding section 30 divides a moving picture compression object signal (YUV420, YUV422), which is a signal obtained by transforming a RGB (Red Green Blue) image outputted from an video signal transforming unit 10 into a MPEG format, into macro blocks of 16*16 pixels and estimates whether motions happened in each macro block, i.e., search area. Then, it obtains a motion vector (VM) through the estimation of motions and estimates information about conditions where motion is out of the search area or where motion can not be expressed by a motion vector. At this time, the motion vector is location information having a most similarity between a current frame image and a previous frame image. By providing such a motion vector, it is possible to perform a compression with a higher compression performance and quality. However, the motion estimation process by the motion estimation unit 31 occupies the most operations when encoding moving pictures. Accordingly, in the case of a mobile communication terminal which requires a low complexity due to a restraint of the CPU performance, it is difficult to embody a function of encoding moving pictures in a software manner because ofthe motion estimation process. In addition, according to the related art moving picture encoding apparatus, in the inter-coding method, each macro block is classified into a block where no motion happened and a block where motions happened, and the block where motions happened is classified into an intra mode and an inter mode depending on its encoding method. The intra mode is a mode of processing the macro block in the same manner as the intra coding, and the inter mode is a mode of encoding the macro block using a difference value between a current frame and a previous frame. The intra mode and the inter mode are distinguished through a difference value between a current frame and the lowest mean value of a macro block unit in the search area which occurs while estimating the motion vector, the lowest mean value being obtained by calculating a difference value between a macro block of a current frame and a macro block of a previous frame. As described above, according to the related art moving picture encoding apparatus using a MPEG, since a previous image restored from compressed image is used when performing an inter-coding, a decoding process is also required when encoding moving pictures. Since inter-coding obtains motion vectors by performing a motion estimation for each macro block and, in addition, performs a DCT and a quantization processes to obtain information about whether motions happened or not in each of the macro block, the inter- coding requires complex operations. Accordingly, it is difficult to embody a moving picture encoding in a mobile communication terminal in a software manner. Further, since a Huffman coding is used as entropy coding, it is not suitable for a real-time service.

DISCLOSURE OF INVENTION Accordingly, the present invention has been made to solve the above-mentioned problems occurring in the related art. The object of the present invention is to embody a moving picture encoding function suitable for a mobile communication terminal which is an application field requiring a real-time service, by applying a Golomb-Rice coding having a low complexity and a fast processing speed as an entropy codrng method when encoding moving pictures. The other object ofthe present invention is to minimize the number of operations, as a decoding process is not required when performing an encoding, by executing a predictive-frame coding using an original image of a previous intra frame when performing an inter-coding for moving pictures. Another object of the present invention is to reduce operations a lot through removing unnecessary processes such as a process for obtaining motion vector when rnter- coding moving pictures, by performing a motion estimation in a macro block unit when performing an inter-coding of moving pictures, estimating whether motions happened using pixel values of a block boundary, and performing an estimation of an intra mode and an inter mode at the same time. The other object of the invention is to save a development cost for addition of a hardware chip allowing encoding of moving pictures suitable for a mobile communication terminal, and to more efficiently provide moving picture service for a mobile communication terminal which is rapidly introduced, by embodying functions of encoding moving pictures in a software manner capable of quickly coping with some conditions, contrary to a hardware chip. In order to accomplish the objects, there is provided a system for encoding moving pictures for a mobile communication terminal having a video signal transforming unit for transforming an inputted RGB image into a moving picture compression object signal having Y, U and V components comprising: an intra-coding section for discrete cosine transforming the moving picture compression object signal, obtaining quantization coefficient by quantizing the discrete cosine transformed moving picture compression object signal, Golomb-Rice coding the quantization coefficient as an entropy coding, and outputting the Golomb-Rice coded quantization coefficient; and an inter-coding section for predictive-frame coding the moving picture compression object signal with the Golomb-Rice coding using an original image of a previous intra frame, estimating whether motions happened in a macro block unit using pixel values of a block boundary and performing an estimation of an intra mode and an inter mode at the same time. Preferably, the inter-coding section may comprise: a motion estimation unit for estimating whether motions happened in a search area, the macro block unit, using an original image of a previous intra frame, for the moving picture compression object signal outputted by the video signal transforming unit; a texture map information storing unit for storing texture map information indicating whether motions happened or not in each macro block; a motion block storing unit for storing a motion block where motions happened, the motion block being outputted by the motion estimation unit; a non-motion block storing unit for storing a non-motion block where no motion happened, the non-motion block being outputted by the motion estimation unit; a discrete cosine transform (DCT) unit for obtaining a DCT coefficient by discrete cosine transforming the motion blocks stored in the motion block storing unit; a quantization unit for producing a quantization coefficient by performing a quantization process corresponding to a quantization width and each frequency component for the obtained DCT coefficient; and a Golomb-Rice coding unit for encoding and then outputting the produced quantization coefficient through a Golomb- Rice coding. Preferably, the motion estimation unit may estimate whether motions happened through a difference between boundary value of a current frame' s macro block and boundary value of a previous frame' s macro block, output respectively the block where motions happened and the block where no motion happened, and output information about whether motions happened or not in each macro block as texture map information. Preferably, the motion estimation unit may perform an estimation of an intra mode and an inter mode, based on a value obtained by squaring a difference value between a macro block of a current frame and a macro block of a previous frame, the difference value being a resultant value obtained when estimating whether motions happened. Differently, there is provided a method of encoding moving pictures for a mobile communication terminal performing an intra-coding and an inter-coding by transforming an inputted RGB image into a moving picture compression object signal comprising: the intra- coding process that discrete cosine transforms the moving picture compression object signal, obtains quantization coefficient by quantizing the discrete cosine transformed moving picture compression object signal, Golomb-Rice codes the quantization coefficient as an entropy coding, and outputs the Golomb-Rice coded quantization coefficient; and the inter-coding process that estimates whether motions happened in a macro block unit for the moving picture compression object signal, classifies the macro block into a motion block and a non-motion block, stores the motion block and the non-motion block separately, encodes and outputs information about whether motions happened or not as texture map information, and performs an estimation of an intra mode and an inter mode for the motion block at the same time when estimating whether motions happened. Preferably, the inter-coding process may comprise: estimating whether motions happened through a difference between boundary value of a current frame's macro block and boundary value of a previous frame's macro block; outputting and storing non-motion blocks where no motion happened and motion blocks where motions happened in each storing unit according to a result ofthe estimation of whether motions happened, and then encoding and outputting texture map information indicating whether motions happened or not in a corresponding macro block; and estimating whether to encode the motion block where motions happened in an intra mode or inter mode at the same time when performing the estimation of whether motions happened or not. Preferably, the inter-coding process may encode motion blocks, estimated as the intra mode, in the same manner as the intra-coding and encode motion blocks, estimated as the inter mode, by encoding different components between an original image of a current frame and a previous image.

BRIEF DESCRIPTION OF THE DRAWINGS The above and other objects, features and advantages ofthe present invention will be more apparent from the following detailed description taken in conjunction with the accompanying drawings, in which: FIG. 1 is a block diagram illustrating structure of a moving picture encoding system according to the related art MPEG mode; and FIG. 2 is a block diagram functionally illustrating structure of a moving picture encoding system for a mobile communication terminal according to an embodiment of the present invention.

**Description ofthe codes at important parts ofthe diagrams**

10: Video signal transforming unit 50: Intra-coding section

51 , 65 : DCT unit 52, 66: Quantization unit

53, 67: Golomb-Rice codrng unit 60: Inter-coding section 61 : Motion estimation unit 62 : Motion block storing unit

63 : Non-motion block storing unit 67: Texture map information storing unit

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, preferred embodiments ofthe present invention will be described with reference to the accompanying drawings. A system of encoding moving pictures according to the present invention is embodied in a software manner, rather than a hardware chip. As illustrated in Figure 2, the system according to an embodiment of the present invention comprises a video signal fransforming unit 10 for transforming an inputted RGB image into a moving picture compression object signal (YUV420) having a MPEG-4 format, an intra-coding section 50 for discrete cosine transforming, quantizing and entropy codrng the moving picture compression object signal with a Golomb-Rice coding, and an inter-coding section 60 for performing a predictive-frame coding with the Golomb-Rice coding, using an original image of a previous intra frame, estimating whether motions happened in a macro block unit using pixel values of a block boundary and performing an estimation of an intra mode and an inter mode at the same time. The intra-coding section 50 comprises a DCT unit 51, a quantization unit 52 and a Golomb-Rice coding unit 53. The DCT unit 51 obtains a DCT coefficient by discrete- cosine transforming the moving picture compression object signal (YUV420) outputted by the video signal fransformrng unit 10. The quantization unit 52 produces a quantization coefficient by performing a quantization process of making the obtained DCT coefficient into a 8*8 quantization matrix corresponding to a quantization width and each frequency component. The Golomb-Rice coding unit 53 encodes and outputs the produced quantization coefficient through the Golomb-Rice coding. The inter-coding section 60 comprises a motion estimation unit 61, a motion block storing unit 62, a non-motion block storing unit 63, a texture map information storing unit 64, a DCT unit 65, a quantization unit 66 and a Golomb-Rice codrng unit 67. The motion estimation unit 61 estimates whether motions happened in a search area, a macro block unit (16* 16), using an original image of a previous intra frame, for the moving picture compression object signal outputted by the video signal fransforrning unit 10. At this time, it estimates whether motions happened through a difference between boundary value of a current frame's macro block and boundary value of a previous frame's macro block, outputs the block where motions happened and the block where no motion happened respectively depending on results of the estimation, and outputs information about whether motions happened in each macro block as texture map information. In addition, it performs an estimation of an infra mode and an inter mode, based on a value obtained by squaring a difference value between a macro block of the current frame and a macro block of the previous frame, the difference value being a resultant value obtained when estimating whether motions happened. The motion block storing unit 62 stores blocks where motions happened. The non-motion block storing unit 63 stores blocks where no motion happened. The texture map information storing unit 64 stores the texture map information outputted by the motion estimation unit 61 , the texture map information indicating whether motions happened or not in each macro block. The DCT unit 65 obtains a DCT coefficient by discrete cosine transforming the motion blocks stored in the motion block storing unit 62. The quantization unit 66 produces a quantization coefficient by performing a quantization process of making the obtained DCT coefficient into a 8*8 quantization matrix corresponding to a quantization width and each frequency component. The Golomb-Rice coding unit 67 encodes and outputs the produced quantization coefficient through the Golomb-Rice codrng. Hereinafter, operations of the system of encoding moving pictures having the above-mentioned structures will be described. First, a RGB image inputted through an inputting device (not illustrated) such as a camera is transformed into a moving picture compression object signal (YUV420) having a MPEG-4 format and then inputted into the intra-coding section 50, by the video signal transforming unit 10. Then, the intra-coding section 50 obtains a DCT coefficient by discrete-cosine transforming the moving picture compression object signal (YUV420) inputted from the video signal fransforming unit 10 and then produces a quantization coefficient by performing a quantization process of making the obtained DCT coefficient into a 8*8 quantization matrix corresponding to a quantization width and each frequency component. After that, the intra-coding section 50 encodes and then outputs the produced quantization coefficient through the Golomb-Rice coding. Accordingly, an encoding for an inputted image frame is made using a spatial correlation in a frame, as a still image. At this time, since a complexity is lower than a Huffman coding, a related art entropy coding method, and a real-time encoding can be made, it can be applied to a real-time service. Meanwhile, the inter-coding section 60 encodes the moving picture compression object signal inputted from the video signal transforming unit 10 using a time relation between a current frame and a previous frame. First, the motion estimation unit 61 of the inter-coding section 60 divides the moving picture compression object signal into macro block units (16*16) and classifies the block units into blocks where motions happened (motion block) and blocks where no motion happened (non-motion block). At this time, it stores information about whether motions happened or not in each block as texture map information. In other words, the motion estimation unit 61 estimates whether motions happened in a search area, a macro block unit, using an original image of a previous intra frame. At this time, it estimates whether motions happened through a difference between boundary value of a current frame's macro block and boundary value of a previous frame's macro block. When it is estimated that a corresponding macro block is a non-motion block where no motion happened, the motion estimation unit 61 outputs the macro block, stores the non-motion block in the non-motion block storing unit 63, and stores information that no motion happened in the macro block in the texture map information storing unit 64. However, when it is estimated that a corresponding macro block is a motion block where motions happened, the motion estimation unit 61 outputs the macro block, stores the motion block in the motion block storing unit 62, and stores information that motions happened in the macro block in the texture map information storing unit 64. Since the information about whether motions happened or not, which is stored in the texture map information storing unit 64, is encoded and outputted, a decoding section (not illustrated) is able to know the information about whether motions happened in each macro block. Accordingly, an encoding process or any subsequent processes is not required for the non-motion block. In other words, the decoding section processes a non- motion block with the texture map information by copying blocks in the previous image. In addition, the motion estimation unit 61 estimates whether to encode the motion block where motions happened in the intra mode or in the inter mode at the same time when it performs the estimation of whether motions happened or not. In other words, blocks can be estimated as a motion block where motions happened, in case that there is movement of objects, change of light or a noise from a moving picture inputting device (for example, a camera). Among the above cases, motion blocks where there is a big movement such as the movement of objects are estimated to be encoded in the intra mode. Motion blocks where there is a small change in images such as minute movements due to lights or noise are estimated to be encoded in the inter mode. The corresponding intra mode encoding or inter mode encoding information is encoded and stored for the decoding section to be aware of it. At this time, in the case of the motion block estimated in the intra mode, block values of an original image are encoded in the same manner as the intra frame coding. In the case of the motion block estimated in the inter mode, different components between an original image and a previous image are encoded.

INDUSTRIAL APPLICABR TY As described above, according to the present invention, when encoding moving pictures, a Golomb-Rice coding having a low complexity and a fast processing speed is applied instead of a Huffman coding as an entropy coding method. Accordingly, it is possible to embody a moving picture encoding function suitable for a mobile communication terminal which is an application requiring a real-time service. According to the present invention, when performing an inter-coding for moving pictures, a predictive-frame coding is performed using an original image of a previous intra frame. Accordingly, since it is not required to use an image restored from compressed previous image as the related art, the number of operations can be minimized. For example, a decoding process is not required when performing an encoding. In addition, according to the present invention, an estimation of whether motions happened is performed in a macro block unit when performing an intra-coding for moving pictures. At this time, it is estimated only whether motions happened using pixel values of a block boundary without obtaining a motion vector, as with the related art. Accordingly, since a process for obtaining a motion vector, and a discrete-cosine transformation and a quantization process required for obtaining information about whether motions happened are omitted, many operations can be reduced. Further, according to the present invention, when performing inter-coding for moving pictures, a motion estimation and an estimation of an intra mode and an inter mode are performed at the same time, rather than performing an estimation ofthe infra mode and the inter mode after obtaining a motion vector, as with the related art. Accordingly, it is possible to reduce the operations of the previously described processes, compared to the related art apparatus of encoding moving pictures. Additionally, according to the present invention, the above-mentioned moving picture encoding functions are embodied in a software manner. Accordingly, it is possible to reduce a burden on additional expenses for developing a mobile communication terminal, which expenses occur due to an addition of a hardware chip for encoding moving pictures. In addition, contrary to a hardware chip, since the functions are embodied in a software manner to simply and quickly cope with some conditions, a moving picture service on a mobile communication terminal can be more effectively provided. While the invention has been illustrated and described with reference to certain preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope ofthe invention as defined by the appended claims.

Claims

WHAT IS CLAIMED IS: 1. A system of encoding moving pictures for a mobile communication terminal having a video signal transforming unit for transforming an inputted RGB image into a moving picture compression object signal having Y, U and V components comprising: an intra-coding section for discrete cosine transforming the moving picture compression object signal, obtaining quantization coefficient by quantizing the discrete cosine transformed moving picture compression object signal, Golomb-Rice codrng the quantization coefficient as an entropy coding, and outputting the Golomb-Rice coded quantization coefficient; and an inter-coding section for predictive-frame coding the moving picture compression object signal with the Golomb-Rice coding using an original image of a previous intra frame, estimating whether motions happened in a macro block unit using pixel values of a block boundary and performing an estimation of an intra mode and an inter mode at the same time.

2. The system according to claim 1, wherein the inter-coding section comprises: a motion estimation unit for estimating whether motions happened in a search area, the macro block unit, using an original image of a previous intra frame, for the moving picture compression object signal outputted by the video signal fransforming unit; a texture map information storing unit for storing texture map information indicating whether motions happened or not in each macro block; a motion block storing unit for storing a motion block where motions happened, the motion block being outputted by the motion estimation unit; a non-motion block storing unit for storing a non-motion block where no motion happened, the non-motion block being outputted by the motion estimation unit; a discrete cosine transform (DCT) unit for obtaining a DCT coefficient by discrete cosine transforming the motion blocks stored in the motion block storing unit; a quantization unit for producing a quantization coefficient by performing a quantization process corresponding to a quantization width and each frequency component for the obtained DCT coefficient; and a Golomb-Rice coding unit for encoding and then outputting the produced quantization coefficient through a Golomb-Rice coding.

3. The system according to claim 2, wherein the motion estimation unit estimates whether motions happened through a difference between boundary value of a current frame' s macro block and boundary value of a previous frame' s macro block, outputs respectively the block where motions happened and the block where no motion happened, and outputs information about whether motions happened or not in each macro block as texture map information.

4. The system according to claim 2, wherein the motion estimation unit performs an estimation of an intra mode and an inter mode, based on a value obtained by squaring a difference value between a macro block of a current frame and a macro block of a previous frame, the difference value being a resultant value obtained when estimating whether motions happened.

5. A method of encoding moving pictures for a mobile communication terminal performing an intra-coding and an inter-coding by transforming an inputted RGB image into a moving picture compression object signal comprising: the intra-coding process that discrete cosine transforms the moving picture compression object signal, obtains quantization coefficient by quantizing the discrete cosine transformed moving picture compression object signal, Golomb-Rice codes the quantization coefficient as an entropy coding, and outputs the Golomb-Rice coded quantization coefficient; and the inter-coding process that estimates whether motions happened in a macro block unit for the moving picture compression object signal, classifies the macro block into a motion block and a non-motion block, stores the motion block and the non-motion block separately, encodes and outputs information about whether motions happened or not as texture map information, and performs an estimation of an infra mode and an inter mode for the motion block at the same time when estimating whether motions happened.

6. The method according to claim 5, wherein the inter-coding process comprises: estimating whether motions happened through a difference between boundary value of a current frame's macro block and boundary value of a previous frame's macro block; outputting and storing non-motion blocks where no motion happened and motion blocks where motions happened in each storing unit according to a result ofthe estimation of whether motions happened, and then encoding and outputting texture map information indicating whether motions happened or not in a corresponding macro block; and estimating whether to encode the motion block where motions happened in an intra mode or inter mode at the same time when performing the estimation of whether motions happened or not.

7. The method according to claim 5, wherein the inter-coding process encodes motion blocks, estimated as the intra mode, in the same manner as the intra-coding and encodes motion blocks, estimated as the inter mode, by encoding different components between an original image of a current frame and a previous image.