Date Added: Jul 2012
Content-Based Video Retrieval (CBVR) is an emerging, challenging and important issue, recently. CBVR involves processing video frames to find the similarity with respect to query video or frame. This paper presents a combined algorithm for detecting and retrieving similar frames in a video from database. The developed algorithm employs a two method to achieve video frame retrieval: scene change detection to identify the shots in the video and similar frame identification. Scene change detection is accomplished using three shot detection algorithms such as Bit plane Triangle Template Matching (BTTM), Genetic based Edge Histogram (GEH) and H.264 motion vector based shot change detection. Once the scene change frames are detected, frame retrieval is achieved using image segmentation and feature extraction algorithms.