摘要
为从视频序列提取运动矢量,进行有效的视频检索,对现存的运动特征的提取算法存在的问题进行了分析,提出了基于自适应块匹配搜索的全局运动特征提取方法。实验表明该算法在大大提高搜索速度的同时,能够获得几乎与全搜索算法相同的峰值信噪比(PSNR)。
In order to extract motion vector to realize video retrieval efficiently, the existing problems and characters of video information were analyzed. Then a global motion feature extraction scheme based on adaptive block matching fast searching algorithm was put forward. Experiments show that the algorithm can obtain almost the same PSNR as the full search algorithm at very low computation cost.
出处
《计算机应用》
CSCD
北大核心
2006年第4期797-798,802,共3页
journal of Computer Applications
基金
广东省科技计划资助项目(2004B16001149)
