Multiple Grey-scale Feature Based Shot Boundary Detection

Effective automatic tools to categorize the searching and retrieval of relevant video materials from the huge data set become vital due to the profusion of video data. Shot Boundary Detection (SBD) is an important preprocessing step of many video processing applications. Though there are many solutions developed in this area, still there is space for improvement. A variety of shot boundaries and different variant features are the big challenges in this area. In this paper, a two-stage SBD method with the combination of Gray Level Co-occurrence Matrix (GLCM), histogram and SIFT Keypoint matching is proposed. The first stage aims to reduce the miss rate and the second stage aims to reduce the false positives. Comparatively better result than the individual feature method is achieved.