一种新的基于运动矢量场及弹性模板的自适应快速搜索算法

A New Adaptive Motion Search with Elastic Pattern Based on Motion Vector Field

该文提出了一种新的基于运动矢量场及弹性模板自适应快速搜索算法 ,它是以视频运动具有高度时空相关性为基础 ,从运动矢量场的均匀性出发 ,综合采用了快速块匹配、自适应弹性模板选择策略、自适应阈值与搜索中止准则等一系列技术 ,是一种具有伸缩性结构的自适应快速运动估计算法 .实验结果表明 ,该算法以极小的搜索代价得到了与全搜索相当的效果 ,并在搜索速度和搜索效果方面明显优于MPEG 4最新推荐的快速运动估计算法 ,特别是在中低码率压缩时有着更优良的性能 ,非常适合于低码率实时视频压缩领域的应用 .

A new adaptive motion search algorithm with elastic pattern based on motion vector field is presented in this paper. Based on high spatial-temporal relativity of the video sequences, the algorithm started with the uniformity of motion vector field, combining a series of skills such as fast block matching,selective strategy of adaptive elastic pattern,adaptive threshold and effective half-stop criteria. It is an adaptive fast motion estimation algorithm with scalable structure. The experiment result shows that the algorithm could obtain almost the same PSNR as the full search algorithm at very low computation cost. Meanwhile, in terms of speed and quality, it apparently excels the newest fast motion estimation algorithm recommended by MPEG-4 Part 7. It has much better capability in low and medium bit-rate compression, thus, it is very suitable to be applied to the field of low-bit rate real-time video compression.