Title :
Block Motion Model for Optical Flow with Smoothness Prior Function
Author :
Tandjung, Stephanus Surijadarma ; Soon, Seah Hock ; Kemao, Qian
Author_Institution :
BeyondLSI, Inc., Tokyo
Abstract :
An explicit constraint is introduced into the Lucas-Kanade gradient based block motion model. This constraint helps the estimation process to consider surrounding motion vectors during its calculation. Consequently a better motion field can be produced than that by the original block motion model. Further, a discontinuity adaptive function is introduced into the Lucas-Kanade equation, which helps to preserve the discontinuities of motion fields
Keywords :
Markov processes; image sequences; maximum likelihood estimation; motion estimation; Lucas-Kanade equation; Lucas-Kanade gradient; Markov random field; block motion model; discontinuity adaptive function; maximum a posterior probability; motion estimation; motion field discontinuity; motion vectors; optical flow; smoothness prior function; Entropy; Equations; Image motion analysis; Image sequences; Markov random fields; Motion control; Motion estimation; Motion measurement; PSNR; Markov random field; maximum a posterior probability; motion estimation; optical flow;
Conference_Titel :
Control, Automation, Robotics and Vision, 2006. ICARCV '06. 9th International Conference on
Conference_Location :
Singapore
Print_ISBN :
1-4244-0341-3
Electronic_ISBN :
1-4214-042-1
DOI :
10.1109/ICARCV.2006.345246
