论文部分内容阅读
本文给出了一种用于MPEG2MP @ML视频编码器单片集成的低功耗运动估值器实现 ,它基于一种低复杂性的改进的多分辨率望远镜搜索算法 ,在粗分辨率运动矢量搜索过程中 ,采用以运动跟踪搜索和自适应望远镜搜索组成的两阶段可提前中断搜索来替换单一的传统望远镜搜索 .模拟结果表明新算法的计算负载仅为传统算法的 30 %左右 ,却可以保持与之相同的视频解码图质量 .本文采用心动阵列和改进的树结构阵列来分别加速运动矢量的粗分辨率搜索和优化搜索 ,并通过门控时钟和屏蔽操作数等方法来实现低功耗 .
This paper presents a low power consumption motion estimator for MPEG2MP @ML video encoder monolithic integration. It is based on a low complexity and improved multiresolution telescope search algorithm. In coarse resolution motion vector During the search, a single traditional telescope search was replaced by a two-stage pre-emptive break search consisting of motion tracking search and adaptive telescope search.The simulation results show that the new algorithm can only maintain about 30% of the computational load of traditional algorithms The same quality of the video decoding picture.In this paper, the cardiac motion array and the improved tree structure array are used to accelerate the coarse resolution search and the optimized search of the motion vector respectively, and the low power consumption is achieved by means of gating clock and shielding operands.