DocumentCode
1136543
Title
Efficient region-based motion estimation and symmetry oriented segmentation for image sequence coding
Author
Cicconi, Pierangela ; Nicolas, Henri
Author_Institution
Signal Process. Lab., Swiss Federal Inst. of Technol., Lausanne, Switzerland
Volume
4
Issue
3
fYear
1994
fDate
6/1/1994 12:00:00 AM
Firstpage
357
Lastpage
364
Abstract
In the paper an efficient coding system for very low bit rates is presented. It is characterized by three operating modes. The intraframe mode combines image segmentation and an efficient luminance representation focalized on the concept of symmetry. The motion compensated mode is based on an adaptive motion representation which provides a good trade-off between the cost of parameter transmission and the quality of prediction. The obtained motion vector field is used for the reconstruction in the predicted mode. Experimental results are given on four test sequences. Simulations show good performance of the coding system both in terms of PSNR and bitrate. In particular, experiments show that the system is capable of working at around 24 kbit/sec with good quality for head-and-shoulder sequences. With more complex sequences, 50 kbit/sec are reached with acceptable degradation of the reconstructed images
Keywords
adaptive systems; image coding; image segmentation; motion estimation; parameter estimation; 24 to 50 kbit/s; adaptive motion representation; degradation; efficient region-based motion estimation; head-and-shoulder sequences; image segmentation; image sequence coding; intraframe mode; luminance representation; motion compensated mode; motion vector field; parameter transmission; performance; predicted mode; prediction; reconstruction; symmetry; symmetry oriented segmentation; test sequence; very low bit rates; Bit rate; Costs; Degradation; Image coding; Image reconstruction; Image segmentation; Image sequences; Motion compensation; Motion estimation; PSNR; Skeleton; Testing;
fLanguage
English
Journal_Title
Circuits and Systems for Video Technology, IEEE Transactions on
Publisher
ieee
ISSN
1051-8215
Type
jour
DOI
10.1109/76.305879
Filename
305879
Link To Document