摘要
镜头边缘检测是实现CBVR的一个重要的步骤.提出了忽略分块极大值的加权直方图的方法,并结合此概念及现有算法分别提出一种改进的窗口法来检测突变和一种改进的基于非相邻帧差法来检测渐变,更加精确地实现了镜头的分割以及对镜头边界点的定位.实验结果表明该算法行之有效.
The shot boundary detection is an important step in realizing CBVR. This article, the method of weighting histogram -neglecting intersected proved window law to detect abrupt detect gradual change with existing the orientation of boundary position change block maximum value is proposed , and it's used to propose an improved , and a improved non-neighbor frame difference based method to algorithms. It has realized more accurately that the partition of shots and of shots. The experimental results indicates that this algorithm is effective.
出处
《辽宁大学学报(自然科学版)》
CAS
2007年第4期304-306,共3页
Journal of Liaoning University:Natural Sciences Edition
关键词
镜头边缘检测
非相邻帧差
忽略分块极大值的加权直方图
shot boundary detection
non-neighbor frame difference
weighting histogram - neglecting intersected block maximum value.