基于RS理论的镜头分割预处理算法

在分析视频数据特征的基础上,引入粗糙集的属性约简理论,提出一种基于RS理论的镜头分割预处理算法.该算法首先从压缩域提取视频帧的DCT系数,在对DCT系数处理后得到DC系数的信息系统,然后利用RS理论对信息系统进行属性约减,最后得到没有冗余的帧序列.实验结果显示,算法处理后,镜头的帧数目大幅度减少,而后续的镜头分割结果基本不变.实验结果表明,此算法在镜头分割前能消除大量的冗余数据,避免了大量的复杂运算.

Preprocessing algorithms of shot segmentation based on RS theory

On the basis of analyzing the video data characteristics,the attribute reduction theory of rough set was introduced and an RS-based shot segmentation preprocessing algorithm was presented,where the DCT coefficients of video frames were first extracted from the compressed-domain and the information system of DC coefficients was obtained,after their processing,then the attribute reduction of the information system was performed with RS theory,and finally the sequence of frame without redundancy was obtained.The experimental results showed that the numbers of the frames decreased to a great extent after the algorithm processing and the subsequent shot segmentation remained basically unchanged.It could be seen that this algorithm could eliminate the redundancy of the data before the shot segmentation,avoiding great amounts of complex computations.