Comparison Of Several Lossless Image Compression And Coding Algorithms

Lossless image compression is mainly applied in some fields where the images are subjected to further processing, repeated compression / decompression and high cost of the image acquiring. It is of significant practical and commercial interest to do research on fast lossless image compression algorithms.This dissertation mainly includes algorithms which present the two development directions of lossless image compression. One is LOCO-I based on adaptive Stat. context module prediction and variable length coding, which is the core of JPEG-LS; the other is the algorithm based on integer wavelet and SPIHT zerotree coding. The context module prediction of LOCO-I is more effective in decreasing the correlation of pixels than traditional prediction. The IB-integer wavelet transform under lifting has low operation complexity, so it fit the real time circuit system which implements the lossless compression algorithm. The SPIHT coding after wavelet transform is of high performance with combinations of DPCM and Huffman entropy coding.In this dissertation, there are two simulations under Matlab software, one is the predictions of JPEG and JPEG-LS, from which it is concluded that JPEG-LS prediction is better than JPEG prediction; the other is using 5/3 and 9/7-M wavelet transforms and SPIHT entropy coding to complete lossless image coding, from the output bit streams and the histograms of wavelet coefficients, it is found that 9/7-M is better than 5/3 in compression rate. The algorithms of 9/7-M wavelet and LOCO-I can both be applied in hardware circuit system to realize real time lossless image compression.