A Novel Lossy Image Compression Based On Color Prediction Technique

ABSTRACT In this paper, a novel lossy image compression technique based on color prediction and Huffman encoding is proposed. The proposed color prediction technique handles the color planes separately and built pixel by pixel through a raster scan order. A color map is generated based on color difference by using the predefined threshold. This color map is encoded by using Huffman encoding. The predicted colors are stored separately as header information along with the encoded stream. Experimental results show that as threshold increases, the compression ratio of the proposed system also increases although PSNR decreases. The proposed method is extensively tested with well known test images. Simulation results show that the proposed compression scheme can achieve a better compression performance.