Algorithm Research On Image Compression Based On Wavelet Transform

The study of the image coding algorithms and its applications is one of the most active areas in the information technology now.Because of the popularization of computer and internet,multimedia entertainment,multimedia communication,digital audio frequency broadcast and high definition television etc,these practical applications enhance the requirement of information data storage and transmission. One of the main problems,which confronted with multimedia technology,the applications and developments of internet,is to solve the denotion,transmission and storage huge image data information.Thus it inevitably becomes one of the internationally recognized hot tasks to explore the efficient image compression coding algorithm.Wavelet transform has become one of the main research directions in image compression coding because of its good time-frequency localization characteristics and match the human visual system.In this paper,we focus on the image compression coding algorithm based on wavelet transform.Firstly,we summarize the basic theory on image coding and image quality evaluation criteria,the development and conditions of image compression technology, the conditions of video coding technology and most international standards of image coding.Then elaborated some key theories about the wavelet image compression, especially on all kinds of wavelet transform,multi-resolution analysis and Mallat algorithm.It also summarize the basic theory of mathematical morphology.On this basis,study the embedded zerotree wavelet coding algorithm and mathematical morphological representation of wavelet data,analyzes the detail process of standards embedded zerotree wavelet coding algorithm,illustrate the morphological representation of wavelet data image coding algorithm.This paper proposed a kind of combination of mathematical morphology embedded zerotree wavelet compression algorithm based on the above research results.The characteristics of this algorithm include:①Separate coding on low-frequency sub-band.②Remove the high-frequency sub-band coefficients.③The important coefficients in the remainder sub-bands,according to different levels of sub-bands to the importance of images reconstructed treatment different, respectively,using a different structuring element templates to realize the important sub-band coefficients of the expansion of mathematical morphology operation.④Proposed new symbols mapping method represent wavelet coefficients.For those descendants of a significant coefficients are all insignificant,the new symbols are used to replace of the successive zero tree roots of frequent appearance.⑤Using the second encoding method.The six symbols after coded with Huffman code use Run-length coding to recode the symbol's binary stream and the quantization of the amplitude.Finally the video image sequence of wavelet compression algorithm were studied,proposed double cross motion estimation in wavelet domain video compression search algorithm based on transform domain anisotropie characteristic, The proposed search algorithm initially uses the multiresolution-spatio-temporal context to find the center of the search,then using double across search algorithm to make motion estimation and compensation according to anistropic characteristic of subband.