Research And Optimization Of Key Algorithm For New Generation Video Coding Standards

With the development of the multimedia and internet technology, digital video coding is widely used in many applications, such as video phone, video conference, IPTV and digital TV. In order to meet customers' increasing demands on video applications, many standardization organizations have enacted many new video standards. The standards of H.264 and AVS, which are called the next generation video standards because of the more excellent performance compared with the traditional ones, are widely studied nowadays.The contents of this paper is a new generation video coding standard (H.264, AVS) key algorithm and optimization, mainly from the transform coding, predictive coding, scalable coding algorithms, such as aspects of research and the principle of the optimization, as well as the coding algorithm ported to DSP platform code-level optimization algorithms.In the transform coding aspects: a generic generating algorithm for n-Order (n=2~k,k>0) integer DCT transform radix is presented. The relationship between amounts of coefficients and order in radix is discovered through research of DCT principle and proven in combination with the nature of cosine function. On this basis, a generic generating algorithm for n-Order (n=2~k,k>0) integer DCT transform radix is presented, which we do not need to analyze floating radix corresponding to integer's. Through rearrange variations of coefficient, the mid-polynomials are extremely regularity. The group of polynomials in arbitrary N-variable is resolved by design a N-digits with M as radix implementing N-loops to exhaust all possible solutions. The experimental results show that the algorithm can find all available radix for arbitrary n×n (n=2~k,k>0) integer DCT as long as the computing capacity is enough.In predictive coding aspects: First of all, an analysis of predictive coding in the intra-prediction and inter-frame prediction process to identify possible improvements which optimize the point. Through a lot of reading and practice, based on inter-frame prediction and intra prediction module are widely applicable to the issue of all zero block threshold is studied in depth. To reduce the computation of H.264 encoding, we proposed some new thresholds which derived from statistics and mathematic methods. When we use these thresholds in motion estimation, simulation results show that the early termination algorithm method achieves approximately a 10%-18% computational saving and about 0.04db PSNR gain, compared to Yang's method. In intra prediction we proposed a mode matrix to reduce the mode decision. Experimental results demonstrate that when we use full matrix in the proposed methods can achieve around 9%-15% improvement in average speedup with no any video quality degradation. If we use simple matrix in the proposed method, we can reduce the entire I frame coding time up to 48% with no significant loss of video quality.In the scalable coding aspects: a new algorithm of scalable video coding based on frequency spectrum is presented. Data partition in MPEG-2 is an early scalable video coding based on spectrum, but it is rough in scalable granularity and low in practicability, while the algorithm raised in this paper can achieve finer granularity of scalability (9 levels). The algorithm base on frequency spectrum has flexible scalability, low complexity and high efficiency, but it brings a problem of error accumulation through prediction. Because if it only transmits part of DCT coefficients, it makes reference frames of encoder differ from reference frames of decoder which results in error accumulation. This problem is solved by setting references frames in transform region in this paper. AVS-P2 is a new generation of video coding standard with independent intellectual property in China, and it has become a national standard. In this paper, the algorithm of scalable video coding is combined with AVS-P2 and is realized under the framework of AVS-P2. Finally, the algorithm was realized based on the software codec model version rm52j of AVS-P2 and the practical results were given.In the code-level optimization: We successfully transplanted AVS coding algorithm to TMS320DM6446 platform based on DaVinci technology. Coding algorithm and project-level optimization, Cache performance optimization, optimization of data transmission (EDMA), compilation of linear optimization of a series of optimization techniques, the performance-related test are given at every stage. Optimization of the final results of this paper: the performance of algorithms without loss of image quality under the premise of the frame rate from the initial post-transplant to the 3.8fps of 93.6fps, to achieve real-time CIF.