Study On Video Compression Algorithm Under Low Bit Rate

Digital video compression coding technique is the key of multimedia communication. The current international standards on video compression have already satisfied the requirements for middle or high bit rate and turned into practicality. However, the existing transfer and store capacity requires more for low bit rate communication. How to compression more efficiently is still a hot spot.This paper is aimed to research the video compression algorithm under low bit rate. The common methods used in video compression at present are introduced and their advantages or disadvantages are analyzed. Considering the relativity of the inter-frame and intra-frame in video sequence, this paper regards the video data as a three dimensional image, transforms them using wavelet transformation and then carries quantized coding and fractal coding. This algorithm not only avoids the motion compensation and enhances coding efficiency, but also increases the compression ratio.In order to improve the compression quality, the VQ (vector quantization) algorithm is researched. Evolutionary strategy and human vision characteristic are introduced into SOFM (Self Organized Feature Mapping) Neural Network to training the codebook of VQ to improve the quantization performance. Next, the mapping density image of every frame can be obtained using the above codebook. Feature vector can be extracted to form distance matrix and video shot can be segmented to reduce the computation of wavelet transformation, at the same time ensure the similarity of videocontent. Finally color system transformation is carried in the segmented video shot before 3D wavelet transformation to distill more redundancy. The transformed low frequency coefficients are coded using vector quantization and high frequency components are coded using quad tree fractal coding, which makes full use of the similarity of wavelet coefficients and overcomes the coding delay caused by the searching in fractal coding. This paper also realizes the decoding system.The experimental results indicate this video compression algorithm has high compression ratio and coding efficiency, good reconstructed quality under low bit rate, also has the advantage of not being restricted by video contents, which has certain practical significance.