NEAR-LOSSLESS IMAGE COMPRESSION BASED ON WAVELET AND HISTOGRAM PACKING

In this paper a Near-lossless image compression method based on wavelet and histogram packing is proposed. Our technique consists of three stages, the first stage is wavelet coding/decoding, second stage is histogram packing and third stage is arithmetic coding. In wavelet stage we used A SPIHT technique with wavelet type biror4.4 and 8 levels. The set partitioning in hierarchical trees (SPIHT) algorithm is an efficient wavelet-based progressive image-compression technique, designed to minimize the mean-squared error (MSE) between the original and decoded imagery. From this stage we will have the residual in pixel domain to be an input to the next stage.In the second stage we try to find the best representation of wavelet residuals that give the minimum entropy that can be achieved using histogram packing and Dijkstra algorithm at a given S (allowed grey level loss). Depending on the fact, the reconstructed near lossless image can differ from the original one within a pixelwise error tolerance. This property is used to convert the histogram of the wavelet residual from the first stage, to a new histogram which is proved to have minimum entropy using Dijkstra algorithm. Hence, new residuals matrix is formed which has minimum entropy and high spatial correlation among its pixels and can efficiently be compressed. Finally at the third stage we used arithmetic coding to encode the new residuals that have minimum entropy.By our proposed algorithm we achieved the impressive PSNR and the minimum entropy that can be achieved at a given required delta loss. In addition to that we achieve an improvement in the compression ratio by 11% than the JPEG-LS