Video shot boundary detection and key frame extraction for video retrieval

Abstract

Shot boundary detection and key frame extraction are the primary steps for video indexing, summarization, and retrieval. We have proposed an advanced and sophisticated scale-invariant feature transform (SIFT) key point matching algorithm. This approach is based on capturing the changing statistics of different kinds of shot boundaries using SIFT, followed by extracting key frames from each segmented shot by means of image information entropy technique. We can enhance the performance of this algorithm by eliminating the redundant key frames by using the edge matching rate technique. The proposed algorithms have been applied to different categories of videos to detect shot boundaries and key frame extraction. The experimental results have been excellent and they show that these algorithms are effective and efficient in performance in terms of detection of shots and extraction of ultimate key frames.