Content-Based Video Indexing and Retrieval using Key frames Texture, Edge and Motion Features

In this paper, a novel algorithm for content-based video indexing and retrieval using key-frames texture, edge, and motion features is presented. The algorithm extracts key frames from a video using k-means clustering based method, followed by extraction of texture, edge, and motion features to represent a video with the feature vector. The algorithm is evaluated on a database of three hundred and thirty five videos (collected from TRECVID 2005, Google, and BBC) of four types. The performance of the proposed framework is compared with volume local binary patterns (VLBP) method. The proposed algorithm outperforms well compare to VLBP method.