| Video is an important media for entertainment, education, communication, and research. Organizing and managing video data is a difficult task mainly due to the relatively large file size required to store the data. There is a need to develop indexing techniques for video data.; Recently, the MPEG compression format has been widely accepted as the international compression standard for digital video. The MPEG format will be the primary way to store and deliver the compressed digital video data.; Using encoded parameters in the MPEG compressed video, this dissertation describes several indexing techniques for scene change detection that are based on the P- and B-pictures in the MPEG video. It also includes an adaptive thresholding algorithm to automate the detection process and improve the detection rate. This thesis also describes an encoding algorithm for frame-type optimization which is based on scene changes in video. This algorithm improves the compression ratio by 35% without loss of quality by optimizing the arrangement of frame types.; This optimized and indexed digital video can be decomposed into a set of video clips based on scene change detected in the video. The user can then randomly retrieve any of video clips based on the detected indices in the video sequence. This optimized and indexed digital video could be useful for education, news and sports broadcasting, and video editing in multimedia development. |
