Proceedings

algorithm analysis angle applications approach approximation assumed boundary changes classification complex components Computer connected consider consists constraints contains contour coordinates corresponding curvature curve defined depth derived described detection determine developed direction distance edge elements equation error estimate example extracted faces field Figure flow function given graph IEEE initial input iteration labels linear matching mean measure method motion node normal object observed obtained operator optical orientation pairs parallel parameters path pattern performed pixel plane points position possible presented problem procedure projection range recognition reference region representation represented respectively rotation sample scene segments selected shape shown shows simple smooth solution space step structure surface Table technique texture tion transform tree values vector vertices vision zero