A fast mode decision method for intra prediction in AVS-M video coding

Abstract

A new method is put forward to decrease the high computational complexity during choosing the most suitable predicting mode in intra-prediction of AVS-M. In-depth studying the fast mode decision algorithm for intra prediction and characteristics of image pixels, the gray entropy is used to describe the feature of gray values in 4×4 block. For determining the candidate modes, the simple SAD function is presented to calculate the sampling points selected by gray entropy in the block. Then only candidate modes will be calculated in RDO function which can achieve the fast mode decision for intra prediction. Simulation results show that the algorithm can decrease 25%-33% of the total time in entire intra coding while maintaining PSNR and bit rate almost unchanged.