الفهرس | Only 14 pages are availabe for public view |
Abstract Video compression is essential for reducing the size of the digitized video files; therefore, they could be conveniently exchanged over networks or stored on optical storage media. Compression results in some changes in the original images forming the video stream and the video pattern itself. Therefore, evaluation of compression performance is an essential component in studying video compression. This thesis starts with a detailed survey of video compression techniques with emphasis on the available standards. These standards are related to JPEG, MPEG and H.261. A number of compression techniques related to MPEG-4 standard have been selected for implementation and evaluation. MPEG-4 employs an object-based video representation, which makes it suitable for video indexing and retrieval. Temporal instances of video objects are referred to as video object planes (VOP), and the shape of the video object is approximated by the coding modes 1 (Intraframe), P (Predictive), and B (Bidirectional). An overview of the MPEG-4 architecture and implementation aspects is given. This thesis concentrates on motion estimation aspects, which are essential for video compression. Motion vector and the different algorithms for motion estimation are studied in details. Block matching procedures are discussed with emphasis on the Full search and the 20-Logarithmic methods. I Performance evaluation is then presented. Criteria chosen are mean square and mean absolute encoding errors, compression ratio, and compression time. Different types of video clips (I00 frames) were used to test the different algorithms. The features of the three chosen clips differed to illustrate the changes in the evaluation criteria. |