An Improved JPEG Compression Scheme Using Human Visual System Model

A perceptual image coder for the compression of monochrome images is presented here, in which the coding structure is coupled with a vision model to produce coded images with an improved visual quality at low bit-rates. The coder is an improvement on the Joint Photographic Experts Group (JPEG -Discrete Cosine Transform (DCT) based) image compression standard, and the structure can be easily extended to implement as an improvement on the new JPEG standard. The proposed coding structure incorporates the human vision model in all stages of compression, and gives very good results compared to the existing JPEG standard. Though the mathematical model used are not new, the simple structure of it which is similar to the coding structure of the JPEG, incorporates the visual processing stages in a very systematic manner in the same way as the visual neurons process the signal. The results presented in this paper reveal that the proposed structure gives very good perceptual quality compared to the JPEG scheme especially as we go for lower bit rates. One of the major advantages of the proposed scheme is that it can be easily extended to a structure in which the rate control optimization can be incorporated, as in the new JPEG scheme.