基于特征的实时镜头切变检测

提出了一种新的基于边缘特征的帧间相似性度量方法，并在此基础上实现了一个实时的镜头切变检测算法。为了降低基于特征的算法的运算复杂度，该算法采用一种快速的边缘模式分类方法从部分解码的码流中提取视频帧的边缘特征，通过考察相邻帧边缘分布的相似性定义了一种反映局部信息的帧间相似性度量。结合反映全局特征的基于彩色直方图的相似性的度量和改进的滑动窗算法，实现了高性能的镜头边缘检测。相对于现有的基于特征的算法，该算法具有更低的运算复杂度，适合实时应用。

Real-time Feature-based Shot Cut Detection

A new real-time shot cut detection algorithm using edge feature is presented. To avoid high computation complexity in traditional feature-based detection algorithms, a fast edge pattern classification scheme working with only partially decoded video information is exploited to extract edge features. Instead of trying to find real edges in a frame, all intensity variations are considered and the ambiguity between them is reduced through separating the edges into eight sets according to their directions and normal directions, so the computation is very fast compared to existing feature-based algorithms, The frame similarity is measured by evaluating the similarity of edge distributions aming consecutive frames. The edge-based measure is combined with color histogram based similarity measure to improve the robustness. An algorithm for detecting shot cuts using an improved sliding window thresholding technique is designed and evaluated. Experimental results show that the algorithm is of high performance and applicable to real-time applications.