An improved laplacian distortion model of Mpeg-4 FGS

Abstract

In this paper, we did research on the relationship between rate and distortion for mpeg-4 FGS enhancement layer. During the course of research on the relationship between rate and distortion of meg-4 FGS enhancement layer, we found that there are big discrepancy between the real data and the RD model based on Laplacian distribution. And what's more, the higher the rate is, the larger the error is. Based on the detailed analysis of quantisation approach used in the mpeg-4 FGS encoder, we put forward an improved distortion model. Experiments' results show that the proposed improved distortion model not only can keep good approximation with the real data when the rate is low, the same as the distortion model based on Laplacian model, but also can keep good approximation with the real empirical data when the rate is high. Our proposed new distortion model has a better performance