自适应宏块编码算法的研究

通过对H.264/AVC宏块编码类型空域相关性的大量实验分析,发现经过率失真优化而确定的宏块编码类型无论是I帧、P帧和B帧,在空间域上都具有极强的相关性。提出了采用区域方差来度量宏块编码类型的空域复杂性,并根据宏块的这些空域复杂性制定出宏块类型预测的准则。依照该预测准则,给出了一种基于宏块类型空域预测的自适应宏块编码快速算法,并在H.264/AVC验证模型JM7.6上实现。大量的实验表明,采用宏块类型空域预测的自适用宏块编码算法,减少了H.264/AVC率失真优化算法的大量无效遍历,在保持图像编码质量和压缩效率基本不变的情况下,视频图像编码速度提高了4%~33%,在一定程度上克服了H.264/AVC率失真优化算法速度慢的缺陷。

Research on Adaptive Macroblock Encoding Algorithm

It is found from a lot of experiments of the spatialdomain correlation of the H.264/AVC macroblock(MB) encoding types that the macroblock types decided by rate-distortion optimization(RDO) are of strong spatial-domain correlation no matter in Frame I,P or B.The region mean square errors(MSE) are used to measure the spatial-domain complexity of MB types and a prediction rule for the MB types is set down consequently.According to the prediction rule,a fast adaptive encoding algorithm is put forward on the basis of MB type spatial-domain prediction.The algorithm is verified by the H.264/AVC test model JM7.6.Extensive experiments show that the modified algorithm reduces the ineffective traversing of the H.264/AVC RDO algorithm considerably.The video-frequency coding time shortens by about 4%~33% while the image coding quality and compression performance are maintained fundamentally unchanged.Therefore, the slowness of the H.264/AVC RDO algorithm is compensated to a certain degree.