摘要
本文提出了一种基于综合相似度的二次差分法,综合考虑小波分析和统计量等多种方法对镜头进行检测。在镜头突变检测中考虑了相邻帧的二次差分法;同时,在渐变检测中引入了自适应阈值法,考虑了非相邻帧的二次差分法,避免了人为因素,克服了其他方法仅适用于一类视频序列或几类视频序列的限制,因而具有较好的鲁棒性。实际视频数据的实验结果表明,本文所用方法对镜头边界检测具有很好的效果。
An important step in content-based video retrieval is the temporal segmentation of video. There are some basic algorithms , including the histogram algorithm,the double-thresholding method,which are used to detect abrupt shot changes and gradual shot changes, but they cannot always attain high-precision detection. We propose a twi-difference algorithm for shot boundary detection based on the similarity that integrates the wavelet analysis and statistics of frames. And the algorithm is used to detect the abrupt shot changes and gradual shot changes based on the adjacent frames and the non-adjacent frames. It is proved that the twi-difference algorithm obviously improves the detecting precision.
出处
《计算机工程与科学》
CSCD
2008年第3期40-43,122,共5页
Computer Engineering & Science
基金
南京邮电大学“青蓝”计划资助项目(NY207095)
关键词
镜头边界检测
帧分块
小波变换
似然比
综合相似度
shot boundary detection
frame blocking
wavelet change
ratio rate. integrated similarity
