DocumentCode
773373
Title
Very low-bit-rate segmentation-based video coding using contour and texture prediction
Author
Eryurtlu, F. ; Kondoz, A.M. ; Evans, B.G.
Author_Institution
Centre for Satellite Eng. Res., Surrey Univ., Guildford, UK
Volume
142
Issue
5
fYear
1995
fDate
10/1/1995 12:00:00 AM
Firstpage
253
Lastpage
261
Abstract
The most efficient video coding standard for low bit rates (around 64 kb/s) is the H.261 algorithm recommended by ITU-TS. However, in certain applications such as mobile audiovisual communications and videophone through PSTN, the available transmission bandwidth is very limited. Therefore codecs working at very low bit rates are required. The paper presents a segmentation-based video coding algorithm that can work at rates as low as 10 kb/s. A novel representation of the contour information using a number of control points is proposed to estimate the contour shapes and locations from the previous frame by using the motion information. The texture parameters are also predicted and only the residual values are entropy coded. In addition two novel postprocessing techniques for edge-profile smoothing and jagged-edge rectification are described
Keywords
edge detection; entropy codes; image representation; image segmentation; image sequences; image texture; parameter estimation; prediction theory; smoothing methods; video coding; 10 to 64 kbit/s; codecs; contour locations; contour prediction; contour shapes; control points; edge-profile smoothing; entropy coded; jagged-edge rectification; mobile audiovisual communications; motion information; postprocessing techniques; representation; residual values; texture prediction; very low-bit-rate segmentation-based video coding; videophone;
fLanguage
English
Journal_Title
Vision, Image and Signal Processing, IEE Proceedings -
Publisher
iet
ISSN
1350-245X
Type
jour
DOI
10.1049/ip-vis:19952195
Filename
487784
Link To Document