The Research Of Listless SPIHT Algorithm For Image Coding Based On Wavelet Transform

With the development of technology of multimedia, network and communications, people are using more and more image information in daily life and every walk of life. As the huge quantity of image date is a great disadvantage to storage, transmission and control, image coding is very important.The DWT(discrete wavelet transform) is suitable to image compression because of its character of multi-resolution analysis, which can decompose image into many subbands of different space and frequency. we could get good compression effect if we code the subbands using significant algorithm.A problem that is called extension method will occur when DWT is applied to image date in the process of image coding. This paper analyses several extension methods that are frequently-used and gives a brief introduction to the symmetric extension method. At first we analyse the symmetric type of the output date after linear phase FIR convoluted with symmetric sequence, and then consider the effect of the decimation process on a symmetrical sequence, and finally, we acquire the way how to choose proper extension method.The EZW(the Embedded zero-tree wavelet) algorithm based on the zerotree wavelet image coding technique is an initiatory and oriented coding algorithm in current image coding theory, and SPIHT(The Set Partitioning In Hierarchical Trees) is one of the most efficient methods in its modified algorithms. However, the SPIHT algorithm needs three lists during coding and decoding to store the coordinates of coefficients and subset trees. The three lists become drawbacks for a hardware implementation because a large amount of memory is needed to maintain these lists. For overcoming this drawbacks an algorithm called LZC(Listless Zerotree Coding)was presented. But the LZC's compression effect is worse than the SPIHT's. On the basis of these, this paper presented a modified coding algorithm, which abandons the use of lists and two maps are used instead. The modified algorithm's working memory requirement is about of 1.4% of SPIHT's and 40% of LZC's. The modified algorithm move the two dimensional wavelet transformed image date to a one dimensional array and then to which do sorting pass. During the sorting pass the state of coefficients is obtained in virtue of the information offered by two maps, then coding coefficients in different way accordingly. Coding results show that the modified algorithm's PSNR value outperform LZC's PSNR value that is about of 0.70 dB,and are close to that of SPIHT.