Research On Image Compression Algorithm Based On Wavelet Analysis

Images gradually become one of main contents of communication with the developing of multimedia technology, and image compressing resolving problems of transmitting and storing is the focus of research and has a rapid development from JPEG, MPEG-1 to JPEG-2000, MPEG-4. At present wavelets analysis has replaced Fourier analysis and turned into a primary transforming tool of new coding algorithms. Because of the good time-frequency locality of wavelet overcoming defects of traditional Fourier transform, Algorithms using wavelet analysis dominate present algorithms, such as EZW, SPIHT, EBCOT and JPEG-2000 standard. Now research of algorithm based on wavelet concentrates on the following aspects; how to organize coefficients model to improve the efficiency of coding; how to quantitate coefficients to utilize relativity and make quality of images accorded with HVS; how to organize code-stream to provide multi-quality service and scalable decoding. Organization model of coefficients is proved to be a important factor, especially for a low rate (high compression ratio). Now algorithms organize coefficients into two models: tree structure and block structure, the former uses the relativity between inter sub-bands coefficients while the later uses the relativity between intra sub-bands neighborhood coefficients.A new algorithm combining pre-processing and modified zero-tree is proposed after analyzing the advantages and disadvantages of present excellent algorithms based on tree-structure model. Contrasted with present algorithms based on zero-tree, the new algorithm has defined new signs of significant map and new zero-tree for coding significant maps and taken an operation of truncating quantization results. In addition, this new algorithm possesses pre-processing before wavelet decomposing and post-processing after wavelet reconstructing. The experimental results show the algorithm has improved efficiency of coding. Contrast to EZW and SPIHT, with the same compressing ratio the new one has obvious higher for smooth images, and has a lower elevation of for images without smoothness and with textures. With the increasing of compression ratio the improving of assumes a trend of declining in general. The new algorithm using improved measurements on above is proved to be a useful method.