The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	Editing MPEG2 video in compressed domain
Author:	Egawa, Ren
View Online:	njit-etd2000-083 (xi, 65 pages ~ 5.6 MB pdf)
Department:	Department of Electrical and Computer Engineering
Degree:	Master of Science
Program:	Electrical Engineering
Document Type:	Thesis
Advisory Committee:	Akansu, Ali N. (Committee chair) Shi, Yun Q. (Committee member) Uzun, Necdet (Committee member) Alatan, A. Aydin (Committee member)
Date:	2000-01
Keywords:	Digital video. Video compression. MPEG (video coding standard).
Availability:	Unrestricted
Abstract:	As MPEG2 becomes the heart of most digital video applications, the need to edit an MPEG2 bitstream video in compressed domain has increased. To realize this need, two major technical issues must be resolved: decoder buffer and segment extraction. In the decoder buffer issue, the edited stream must ensure that the buffer would not overflow or underfiow as the result of the editing. This thesis invented a mathematical model that could describe the editing task and predict the editing effect on the buffer. Based on this model, the thesis then built a method to edit a stream safely. This method could produce an edited stream that maintained an MIPEG2 decoder's buffer behavior as if the decoder were receiving a non-edited stream. In the segment extraction issue, the proper method to extract a segment from a source bitstream was discussed. The segment must be independent, that is, no coded picture in the segment would be associated with any other picture data outside of the segment. A proper extraction should also maintain the decoding order and display order of the picture sequence. This thesis visited several past studies on this issue, and proposed a practical method to perform segment extraction. Simulation results on the discussed methods were also presented.