1 Implementation and evaluation of residual color transform for 4 4 4 lossless RGB coding Pooja V Agawane and K R Rao Fellow IEEE Electrical Engineering Department University of Texas at Arlington Arlington Texas Abstract In today s world professional video applications demanding very high video quality are gaining a lot of attention from both academia and industry These professional video applications are highdefinition TV DVD high quality cinema video post processing algorithms studio editing content distribution and many others All these applications demand a very high quality video Along with achieving video of the highest quality it is essential to focus on the coding efficiency This research aims at addressing these real life needs This research implements lossless coding of video in 4 4 4 sampling format This guarantees a high fidelity at the output In order to achieve a high coding gain the red green and blue RGB components of a video signal are decorrelated by transforming them into the newly defined YCgCo color space 27 YCgCo color space was introduced in the Fidelity Range Extensions FRExts amendment 4 of H 264 MPEG 4 part 10 Advanced Video Coding standard 1 Another important concept implemented as a part of this research is the residual color transform RCT 9 RCT was introduced in the High 4 4 4 profile of the FRExts Index Terms H 264 MPEG 4 part 10 AVC lossless coding residual color transform YCgCo color space I INTRODUCTION video signal is represented using three components one luminance and two chrominance or color difference components The human visual system is less sensitive to the chrominance components as compared to the luminance component Hence the chrominance samples can be represented at a lower resolution than the luminance samples This technique is called the chroma subsampling technique 20 Its advantage is that it leads to high compression at the cost of introducing an insignificant distortion in the video output However for professional video applications this distortion is not acceptable 8 Hence the video processing is performed in the 4 4 4 sampling format In this format all the components are represented at equal resolution Due to increasing demand for usage of 4 4 4 sample space there is also a need for efficient coding tools in this space Lossless coding is implemented to address the high video quality demand The lossless coding coupled with 4 4 4 sample space leads to high volumes of data Achieving a good compression ratio along with good coding efficiency is the motivation of this research This research aims at implementing an algorithm to achieve lossless coding for YUV 4 4 4 A sequences High definition sequences 19 are used in this research to address the needs of the professional applications The proposed algorithm implements the conversion of YCbCr to RGB The inter prediction is performed in the RGB domain The residual signals obtained are transformed to YCgCo color space 7 These transformed signals are then encoded using arithmetic coding The stage of quantization is bypassed in the implemented algorithm to account for lossless coding The size of the encoded bitstream is measured and the compression ratio is calculated The coefficients are then reverse transformed to the RGB domain to obtain the residual signals in the RGB domain The reference frame is added to the residual frame to reconstruct the original frame The output frame is displayed and the error frame is calculated The quality of the reconstructed frame is measured using the metrics of mean square error and peak signal to noise ratio II OVERVIEW OF H 264 H 264 MPEG 4 part 10 AVC is one of the latest video coding standards introduced in the world of video compression 1 It was developed by the Joint Video Team JVT consisting of VCEG Video Coding Experts Group of ITU T International Telecommunication Union Telecommunication standardization sector and MPEG Moving Picture Experts Group of ISO IEC 1 This standard is noted for enhanced compression efficiency It can support various interactive video telephony and noninteractive broadcast streaming storage video on demand applications as it facilitates a network friendly video representation 2 The previous coding standards of MPEG1 MPEG 2 MPEG 4 part2 H 261 H 262 and H 263 1 31 are the basis on which the H 264 is developed It uses the basic principles of transform for reduction of spatial correlation quantization for control of bitrate motion compensated prediction for reduction of temporal correlation and entropy coding for reduction in statistical correlation The improved coding efficiency of H 264 has resulted due to changes in the functional elements by including the following enhancements Adaptive intra picture prediction Small block size transform with integer precision Multiple reference pictures and generalized B pictures Variable block sizes Quarter pel precision for motion compensation Content adaptive in loop deblocking filter Improved entropy coding by introduction of CABAC context adaptive binary arithmetic coding and CAVLC context adaptive variable length coding 2 However with the increase in the coding efficiency there is an increase in the complexity to the encoder and the decoder of H 264 To reduce the implementation complexity several techniques like multiplier free integer transform are used In order to develop error resilience for transmission of information over the networks H 264 supports the methods of flexible macroblock ordering switched slice arbitrary slice order redundant slice data partitioning and parameter setting H 264 AVC standard is comprised of a wide range of coding tools Also the standard addresses a large range of bit rates resolutions qualities applications and services Not all the tools and all the bitrates are required for any given application at a given point of time All the various tools of the H 264 are grouped in so called Profiles to maximize the interoperability while limiting the complexity 5 Also the various levels define the parameters like size of decoded pictures bit rate etc The basic profiles defined for H 264 are baseline profile extended profile and main profile Fig 1 illustrates the coding tools for the various profiles of H 264 Encoder specified scaling matrices for perceptually tuned frequency dependent quantization Residual color transform consisting of a reversible integer based color conversion from 4 4 4 RGB to YCgCo color space Efficient lossless representation of the video
View Full Document
Unlocking...