Human visual system-based rate control algorithm for H.264/AVC

To improve the performance of rate control algorithm G012 for H.264/AVC, a human visual system-based rate control algorithm is proposed in this paper. The frame difference ratio of current frame to its previous frame was utilized to represent the motion complexity of the frame, and to be the guidance for bit allocation at frame level. Moreover, simple and effective scheme for measuring visual sensitivity of macroblock via analyzing the motion information and texture characteristics of the frames was introduced, and the bit allocation at basic unit was based on the measured visual sensitivity. The experimental results show that the rate output by the proposed algorithm is more close to the target rate, and PSNR fluctuation is lower compared with that of G012 algorithm, So the subjective quality of the reconstructed video signals is more satisfying.