基于递推模式的帧内预测快速算法

针对H.264帧内预测中存在的边缘像素预测不精确问题,提出一种基于递推模式的帧内预测算法.利用宏块内部像素之间相关性更强的特点,参考同一块内已经预测过的像素预测其他像素.为了降低H.264帧内编码的复杂性,提出一种基于递推模式的帧内预测快速算法.利用4×4块在空域和频域上的特性,采用层次性结构选择需要考察的候选模式,减少帧内预测所要计算的模式数量.根据4×4块与16×16宏块之间的关系,充分利用4×4块的预测结果,快速选择16×16宏块的预测模式.实验结果证明:与以前的快速算法不同,这种基于递推模式的快速算法不仅能在很大程度上减少预测以及重构的计算量,降低帧内编码的计算复杂度,还能提高帧内编码的效率.

Hierarchy-based Fast Intra Prediction Algorithm

In order to solve the problem of the inaccuracy in predicting pixels at the boundary of the block in H.264 intra prediction, a hierarchy-based intra prediction algorithm was presented in this paper. Because of the close dependency of pixels within marco block, the predicted pixels within the same block were taken as references to predict other pixels. A hierarchy-based fast intra prediction algorithm was presented to reduce the computational complexity of intra coding in H. 264, which adopted candidate modes required for hierarchy-based structure by utilizing characteristics of 4×4 block in spatial and frequency domain to reduce the quantity of modes for computation. Furthermore, the relationship between 4×4 mode and 16×16 mode was exploited in fast algorithm. The test results indicate that other than previous fast algorithms, hierarchy-based fast algorithm can not only decrease the computational complexity in intra coding, but also, more important, improve coding efficiency.