一种基于特征跟踪的视频镜头检测算法  

Video shot boundary detection algorithm based on feature tracking

作　　者：高飞[1] 史阳[2] 

机构地区：[1]广东深圳大学信息工程学院,广东深圳518060 [2]西安电子科技大学电子工程学院,西安710071

出　　处：《计算机工程与应用》2007年第8期32-36,共5页Computer Engineering and Applications

基　　金：国家自然科学基金(the National Natural Science Foundation of China under Grant No.60202004)

摘　　要：将视频切分为镜头是视频内容分析及基于内容的视频检索和浏览的第一步。针对视频镜头边界检测,提出了一种基于特征跟踪的新算法。该算法从镜头起始帧中提取出一组角点特征,然后在后续帧中基于Kalman滤波进行特征窗跟踪,最后根据对应特征窗内的像素特征变化规律,得到镜头边界检测所需的测度,判断镜头切换的性质和渐变镜头的起止时间。实验结果表明该算法运算复杂度低,且具有较强的鲁棒性。Partitioning a video sequence into shots is the first step toward video-content analysis and content-based video browsing and retrieval.A novel video shot boundary detection algorithm is presented based on the feature tracking.First,the proposed algorithm extracts a set of comer-points as features from the first frame of a shot.Then,based on the Kalman filtering,the extracted features are tracked with window-matching method from the succeed frames.According to the characteristic pattern of pixels value changing within the tracked windows,the measure of shot boundary detection can be obtained to confirm the types of transition and the time interval of each gradual transition.The experimental results illustrate that the proposed algorithm is effective and robust with low computational complexity.

关 键 词：镜头边界检测 角点检测 特征跟踪 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]