Wavelet-based Scalable Video Coding

PSNR, compression ratio and algorithmic complexity are most important factors which can be used to evaluate coding methods of image and video. Balancing those factors to achieve the best compression effect is the main research purpose at all times. Requirements to the compression ratio and algorithmic complexity become higher and higher due to the development of Internet, so that most traditional compression methods of image and video can not satisfy current typical situations. Due to limited abilities of storage and transmission, how to improve the compression ratio and lower the transmission bit rate while guaranteeing the quality of the transmitted image and video is crucial. Based on the research of image coding and video coding methods, this paper presents a new algorithm of compression coding. Li this paper, explored investigations and researches have been taken, which includes three topics:At first, this paper analyzes the theory of wavelet transform and the methods of wavelet lifting, and lifts the (9-3) filter, then analyzes the methods of boundary extended.This paper use the wavelet transform to construct a line scan based set partition coding algorithm, which actually is an embedded algorithm and takes full advantages of the self similarities of different scale coefficients. It can produce the embedded bit stream and realize bit level precision of the bit rate scalability. The feature of SPECK algorithm is inherited by using blocks as basic partition units, which can be executed in parallel mode and has better efficiency. The compression ratio can be improved by eliminating correlations between lines when line scanning is executed in the basic blocks. Our algorithm can improve compression ratio about 15% or so. Of course the quality of the resumed image declines to some extent, but it is not obvious in the context of continuous video playback.In the video coding part of the scheme, this paper presents a new coding method based on intra-frame difference anticipation. In virtue of the related theory of anticipated coding, not only the pixel correlation between adjacent frames but also the pixel correlation in the same frame is taken into account in our algorithm. In the transmission part of our scheme, a method embedding identity codes into the layered bit streams is presented which can control the scalability of the bit stream. In sum, our algorithm has excellent self-adaptation to the bandwidth changes, and has better efficiency of the compression ratio, algorithmic complexity and PSNR. It is easy to realize and can totally fulfill the requirements of the real time video compression and transmission in our monitoring system.Finally, the conclusions are made, some important issues in CBR research are identified, and some promising research directions are given.