Near lossless image compression system

Image compression has become an important issue in relation to storage and transmission. A very simple and efficient near lossless image compression system is suggested in this paper. In the forward compression algorithm, image pixels are shifted right by the base value. Then, Raster, Orthogonal, Diagonal and Snake scanning methods are applied before finding the difference between pixels. Finally, image data is Huffman coded and sent. Various scanning methods exploit redundancy between image pixel values and the method which gives the best compression ratio is chosen. The proposed algorithm provides an average compression ratio (CR) of 3.256 for very high quality reconstructed images. The CR increases as the base value is increased. The average CR and PSNR for base value 16 is 3.628 and 35.1912 respectively. Orthogonal scanning gives better results for most of the test images.