Predictive Lossless Color Image Compression using Arithmetic Operation

This paper describes a simple and efficient method of colour image compression/decompression using the fundamental application of subtraction and formation of a dynamic dictionary after three stages of subtraction. Pixels in different segments generated by a chosen mask window are adjusted such that after three stages of subtraction, either all zero or at most one non-zero element is left. The dictionary stores the coordinates and the value of the non-zero elements. Colour transformation that decorralate colour components allow a significant increase in compression ratio achieved by use of standard lossless compression technique applied independently to the each component (RED, GREEN, BLUE). The experimental results show that this algorithm, using the fundamentals of predictive encoding gives a lossless technique of colour image compression. This approach gives better compression ratio than some other lossless techniques and JPEG, JPEG 2000, JPEG LS etc. and the process have 100% PSNR value.