一种新的使用两种比特分辨率图象的块匹配运动估计算法及其低功耗VLSI结构

本文给出了一种新的块匹配运动估计算法,它根据视频图像内容的复杂程度自适应地选择常规的或者低比特分辨率的图像来进行块匹配,并且采用了一种混合使用两种比特分辨率图像的新望远镜搜索算法.模拟结果表明,新算法具有较低的计算复杂性,并且能够保证较好的视频质量.基于该算法,我们设计了一种新的脉动阵列结构的搜索引擎.该引擎具有可分割的数据通道,从而在使用低比特分辨率图像进行块匹配时能够通过加强处理的并行性来提高吞吐率.新的运动估计器可工作在较低的时钟频率和电源电压之下,具有低的功耗消耗.

New Block-Matching Motion Estimation Algorithm Using Dual-Bit-Resolution Image and Its Low-Power VLSI Architecture

A new motion estimation algorithm is proposed in this paper.It adaptively selects the full or reduced bit resolution image for block matching based on the texture complexity of the matching block.In addition,a new telescopic search method using dual bit resolution image is adopted.Simulation results show that the proposed algorithm has low computational complexity while achieves good image quality.We have designed a novel systolic architecture based search engine that has split data paths.Enhancing the operating parallelism in the low bit resolution case heightens the throughput.The new motion estimator works at a low clock frequency and a low supply voltage,therefore has lower power consumption.