An Image Compression Algorithm Based On Wavelet Transform

Image compression could be achieved based on human visual system and redundancy occurred in image data. In order to improve the rate-in-use of channel and memorizer, the technology of image compression based on wavelets transform which is regarded as one of the main research directions about image compression, with the target of increasing in compression rate and compression speed while obtaining satisfactory image, has very important significance in theory and practice.Lifting scheme is regarded as an efficient method in wavelet transform. Its Invertible transform could be obtained easily upon the base of keeping perfect reconstruction about inputting data. Lifting scheme has been used widely in the field of image compression, meanwhile becomes the general method of construction second era wavelet transform. In the paper, we describe the basis theory as well as the analysis and reconstruction about lifting scheme while review the traditional wavelets theory. Compared the predict/update lifting scheme with update/predict lifting scheme, we introduce the adaptive lifting scheme and adaptive theory.Upon the base of canvassing the adaptive lifting scheme, an improved lifting scheme with multi-layer is proposed in this paper. With the 3 layers' structure of predict , update and adaptive, it improves the performance of low-pass in update and has the ability of selecting algorithm adaptively in allusion to smooth area and edge area of image by designing the predictor in adaptive neatly. By experience, the coefficient's entropy transformed by improved adaptive lifting scheme is better than that of the rest kinds (inadaptable lifting scheme and update/predict adaptive lifting scheme).With limit of visual sense degree, human visual has the character of no sensitization to sharp changes on image's edge and sensitization to lightness of image as well as feeble resolution to colors. The character is great helpful for us to implement the image compression. In this paper, by introducing the human visual system, we issued different weight to wavelets coefficients according to the sense degree of human visual to different area of image. Thus, The most important coefficient could be transmitted first so that the quality of image reconstructed could be improved further more.In the paper, we redefined zero tree structure in Set Partitioning in Hierarchical Trees (SPIHT) algorithm. By using new zero tree structure, we could reduce the length of set chain list named LIP and save RAM; on the other hand, we could reduce the length of set chain list named LIS and catch the insignificant information availably, meanwhile could improve the proportion of significant coefficient in set chain list named LIS and reduce the time of scanning-sorting, as well as increasing the dichotomizing speed of threshold value and more significant coefficient appearing in bit stream early.A new image coding algorithm based on the improved adaptive lifted wavelet transform and human visual system is proposed in this paper. It can be regarded as an improved version of the SPIHT algorithm. Unlike SPIHT, the improved lifted wavelet transform is used for standard image of testing, the zero-tree structure in SPIHT is redefined and the coefficients after wavelet transform are issued suitable weight by taking into account human visual system (HVS) sufficiently in the compression algorithm. The experimental results show that the new image compression algorithm performs better than that of SPIHT in the aspects of recovery image quality, requirement of RAM and coding/decoding time.