一种基于镜头属性的关键帧提取系统

System of Key Frame Extraction Based on Shot Attribute

研究并设计了一种基于镜头属性的关键帧提取系统。该系统将镜头分成定格、拉伸、静态场景和动态场景4大属性,以分析相邻采样帧中若干条线段的灰度分布曲线相似度为基础,并结合了图像灰度直方图的相似度分析,按照不同的镜头属性提取关键帧。该系统计算量小,满足实时性分析的要求,能较好地说明镜头的类别。

The study and design of a key frame extraction system based on shot attribute is described in this paper. The system divides the shot into four attributes, namely static, zoom, static scene and dynamic scene. The system extracts key frames according to different shot attributes based on analyzing the co,Telation of gray distribution curve of the border of two sample in,ages in video stream together with the correlation analysis of gray histogram of images. Less computation and real time analysis is realized in this system and the key frame extracted can represent the shot attribute properly.