基于主颜色和纹理特征的TS模型匹配方法  

An algorithm of TS matching based on dominant color and texture feature

作　　者：冯自星[1] 覃团发[1] 岳晓峰[1] 陈跃波[1] 

机构地区：[1]广西大学计算机与电子信息学院,广西南宁530004

出　　处：《电子技术应用》2012年第6期138-140,共3页Application of Electronic Technique

基　　金：广西科学基金项目(桂科自0991058);广西高校人才小高地建设创新团队资助计划项目(桂教人[2007]71号)

摘　　要：为了弥补视频匹配方法的不足,解决基于运动特征视频匹配的不精确以及基于内容视频匹配的运算复杂度大等问题,提出一种基于主颜色和纹理特征的TS模型匹配方法,在减少时间复杂度和增加配准率基础进一步提高视频匹配的效率。大量实验证明,该方法有效的解决了上述问题,比传统方法相配准率提高约20%,算法时间复杂度降低约45%。In order to overcome the shortcomings of video matching methods, make up for the inaccuracy of video matching based on motion features and improve the large computing complexity of video matching based on content, and so on, this paper will put forward a TS method based on dominant color and texture feature. It will improve the efficiency of video matching, from the perspective of reducing time complexity and increasing the registration rate. By many experiments, this method solves the above problems effectively. Registration rate increases by about 20% than traditional methods, and the time complexity of algorithm reduces by about 45%.

关 键 词：镜头分割 视频匹配 主颜色 纹理特征 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]