A Study Of Still Image Compression Coding Techniques Based On Wavelet Transform

With the widely application of the digital image technology and multimedia technology, the image becomes the main carrier of information transmission in modern society. Also, because image information contains a huge amount of data,great difficulties have brought to information transmission and storage, the solution is image compression. Therefore, to explore efficient image compression coding method has become an internationally recognizing important research focus.The wavelet transformation technology based on its good time frequency partial characteristic and the eye visual characteristic,has obtained widespread application and research in the image code domain.This paper analyzs the image coding relevant knowledge briefly, expounds the wavelet-based image compression technology, and focus on the structure of the encoder, encoding steps, and several key issues which the wavelet encoding involving. It's also analyzs the energy distribution of the wavelet coefficients through the simulation, and then shows the advantage of the wavelet transform using for image compression.This paper deeply analyzes the embedded zerotree wavelet coding algorithm (EZW) and two classical improved algorithm which are based on the ideas of EZW. In this process, this paper emphases on the main features,principle and the realization of the EZW algorithm,and comparatively analyzes the differences between the three algorithms. Proposes a quantization truncation method for quantization results to these three algorithms,and simulation experiments prove this method can improve compression ratio.At the same time,this paper improved some deficiencies of the EZW coding, proposed an improved image coding algorithm based on EZW and HVS. The improving points are the following aspects.Firstly,giving the definition of zero-tree structure anew, to reduce the algorithm execution time. Secondly,re-classifying the significant symbols, on the one hand,considering the features which the human eye is sensitive to edge imformation, one the other hand,reducing the coding symbols which are unnecessary. Thirdly,combining the wavelet zerotree coding with Huffman coding, run length encoding, further improving the compression ratio. At the last of this paper, experiments results prove that, the improved coding method improves the subjective visual quality and the PSNR of the reconstructed image, it is feasible and effective.