Abrupt Shot Detection in Video using Weighted Edge Information

Abstract

Content Based Video Retrieval (CBVR) has been extensively utilized for automatic indexing, retrieval and management of video data. Segmentation of video is the prominent step in Content Based Video Retrieval. In this paper, we focus on automatic detection of abrupt shot cuts in video sequences. The proposed approach exploits the edge information of an image of a video frame for its characterization. A (2x2) mask of sliding window is used in both overlapping and non overlapping mode to assign binary weights to the edge information of an image. The binary weights evaluated for each mask is used to construct histogram for each image which forms the feature vector to represent an image. The Euclidean distance between the feature vectors of adjacent frames of a video are computed and these values are used for shot cut detection process using adaptive thresholding. To check the efficacy of the proposed shot boundary detection approach, experiments were carried out on a subset of standard video data set TRECVID 2001. The experimental results obtained by the proposed algorithm outperform some of the existing shot boundary detection algorithms in terms of precision, recall and F-measure rates.