Improved Macroblock-Level Rate Control Algorithm with Visual Properties

Paper presents an improved macroblock level rate control scheme for H.264/AVC based on Visual Properties and spatial-temporal correlations. An effective segmentation method is proposed to extract the motion regions, which directly control bit allocation among different regions so as to achieve better visual quality in the regions of interest. By sufficiently analyzing the spatial and temporal correlation, improve the existing MB level algorithm, such as enhance the accuracy of mean absolute difference estimation and the prediction header information, improve the data point selection for rate-distortion model and bits allocation based on the linear distortion model. Experiment results show that compared with JVT-G012, our new scheme can improve the average PSNR up to 0.3dB, higher the coding efficiency and improve the subjective visual quality.