H.264/AVC video coding standard can achieve roughly half of the bit-savings over MPEG2 and MPEG4 for a given quality. However, this comes at a cost in considerably increased complexity at the encoder and thus increases the difficulty in hardware implementation. The high redundancy that exists between the successive frames of a video sequence makes it possible to achieve a high data compression ratio. Motion estimation (ME) plays an important role in motion compensated video coding. A fast motion estimation algorithm for H.264/AVC is proposed based on centered prediction, called centered prediction based fast mixed search algorithm (CPFMS). It makes use of the spatial and temporal correlation in motion vector (MV) fields and feature of all-zero blocks to accelerate the searching process. With the initialized searching point prediction, adaptive search window changing and searching direction decision, CPFMS is provided to reduce computation in block-matching process. The experimental results show that the speed of CPFMS is nearly 12 times of FS with a negligible peak signal-noise ratio (PSNR) loss. Also, the efficiency of CPFMS outperforms some popular fast algorithms such as hybrid unsymmetrical cross multi-hexagongrid search and a novel multidirectional gradient descent search evidently.
新一代的高性能视频编码(HEVC)通过应用各种先进技术来大幅提高视频编码的性能,然而,这些方法也大大增加了整个编码过程,尤其是预测编码阶段的计算复杂度。该文提出一种应用于高性能视频编码帧间预测的快速单元划分算法。该算法基于HEVC灵活的四叉树结构,根据被编码单元的运动特性来确定所编码信息的阈值,并以此作为单元划分的提前终止条件,从而减少了不必要的单元划分操作和率失真代价的相关计算,达到节省编码时间和降低编码复杂度的目的。实验结果表明,在峰值信噪比(PSNR)损失仅为0.0418 d B的情况下,所提算法可以平均降低46.1%的编码时间。
