基于运动信息分布的帧间预测模式选择算法

Inter-frame Prediction Mode Selection Algorithm Based on Motion Information Distribution

多模式预测可以提高H.264标准的编码效率和率失真性能,但会增加编码复杂度。针对上述情况,分析多预测模式帧间预测的复杂度,提出一种基于运动信息分布的帧间预测模式选择算法,以宏块中全零子块的分布情况、各子块的绝对误差和及运动向量的大小作为缩小待选模式集的判据,从而提高编码速度。实验结果表明,该算法的编码质量与全搜索算法相当,编码速度提高了37%～93%。

Multi-mode prediction can improve compression efficiency and rate-distortion performance of H.264,but it also increases the coding complexity.This paper proposes a mode selection algorithm according to the complexity of the variable block-size in motion prediction.The all-zero block distribution in the macroblock,Sum of Absolute Difference（SAD） of the blocks in macroblock and volume of motion vector for macroblock are utilized as decision criterions to lessen candidate mode set and to enhance the coding speed.Simulation result demonstrates that the algorithm can get 37%~93% enhancement of encoding speed versus full search algorithm while incurring imperceptible image quality.