Estimation of Motion Vector Fields

Rasmus Larsen

AbstractThis paper presents an approach to the estimation of 2-D motion vector fields from time varying image sequences. We use a piecewise smooth model based on coupled vector/binary Markov random fields. We find the maximum a posteriori solution by simulated annealing. The algorithm generate sample fields by means of stochastic relaxation implemented via the Gibbs sampler.
KeywordsMarkov random fields, optical flow
TypeConference paper [Without referee]
ConferenceProceedings of the 2nd Danish Conference on Pattern Recognition and Image Analysis (DANKOMB, DSAGM yearly meeting), Aug. 25-26
EditorsPeter Johansen
Year1993    pp. 37-42
PublisherDept. of Computer Science (DIKU), University of Copenhagen
Address
SeriesDIKU Rapport 93/17
ISBN / ISSNISSN 0107-8283
Electronic version(s)[pdf]
BibTeX data [bibtex]
IMM Group(s)Image Analysis & Computer Graphics