Title :
Computation of optical flow by matching edgels
Author_Institution :
Hirst Res. Centre, GEC Res. Ltd., Wembley, UK
Abstract :
A method for computing optical flow at intensity edges is presented and discussed. A relaxation algorithm is used, with neighbourhood support including the disparity gradient limit constraint. The flow vectors are decomposed into principal components, along which directions the errors are extrema. The algorithms described are used for the computation of optical flow in ISOR. ISOR is a set of programs designed by D.W. Murray, written to determine the identity, pose and motion of a geometric object from a sequence of images of the object in motion. The form of the constraint function is considered and a comparison of results obtained using different forms of the compatibility function is given, using both synthetic and real imagery
Keywords :
computer vision; ISOR; compatibility function; computer vision; disparity gradient limit constraint; edgels matching; geometric object; intensity edges; machine vision; optical flow; relaxation algorithm;
Conference_Titel :
Motion and Stereopsis in Machine Vision, IEE Colloquium on
Conference_Location :
London