Research Of Wavelet-based Low Bit Rate Video Compression

Digital video has large amount of data which makes video storage and transportation difficult. Digital video coding is one of the key technologies to resolve this problem. The several international standards of H.26x and MPEG are promulgated by ITU-T and MPEG(Moving Picture Experts Group) based on summarizing the latest and efficient research achievements. The standards have recommended the endeavor direction of video compression research works.The coded image using DCT has block effect , and the compression efficiency is not good. While wavelet transform is suitable in image compression due to its multiresolution or multiscale analysis property. This paper mainly study low bit rates video coding based on wavelet transform. Two key technologies of video compression based on wavelet transform are researched : (1) wavelet coefficient encoding. Embedded Zerotree Wavelet Encoding (EZW) algorithm is the most efficient method coding the wavelet coefficient. This paper presents an improved EZW (IEZW) algorithm due to the lack of EZW. The algorithm improve the coding rate with little memory requirement ; (2) Motion estimation search algorithm. Motion compensation is used to cut the redundance among frames, and the performance of Motion compensation is decided by search algorithm. Some well-known fast search algorithms are systematically analyzed, and a new search algorithm---prediction diamond search (PDS) algorithm is proposed. This algorithm improves search performance and reduces search steps compared with Diamond Search by prediction motion vector.A wavelet-based low bit rates video compression algorithm is then introduced and both the coder and decoder are implemented by software. This coding algorithm can compress video by the method of combining inter-frame compression and intra-frame compression. First the residual frame is computed after processing video sequence by motion compensating. Then this residual frame is wavelet transformed using lifting algorithm. Finally the wavelet coefficient is coded using IEZW algorithm and arithmetic coding . Moreover the coder and decoder are realized by C/C++ language. The experimental results show that this method has better compression performance and faster processing rate without the block effect that may appear in image compression algorithm based on DCT. This algorithm also can compress video at random bit rates. The coder and decoder can be used in video telephone , remote monitering and so on after they are optimized and reinforced.