Curve Matching with Probabilistic Relaxation

Reliable curve matching is a difficult proble~rl but is required in Inany visio11-based applicatio~ls. It is particularly difficult when the edges in question are not li~rlited to be on straight lines. Our 1riai11 corltribution in this paper is to propose a new algorithr~i for curve matching (illcludi~lg lines). The nlethod is described within a probabilistic relaxatio~l fra~rlework. Novel si~~lilarity-invariant urlary and bi~lary l~ieasure~rlents suitable for curves are developed, and a11 additiorlal r~leasurerrierlt is irltroduced to 111odel the u~lcertairity of tlie binary ~neasure~nents, which is very i~~lpor ta~l t i11 corrlputi~ig the matching support fro111 neighboring matches. Experinlents with co~rlplex real scenes show that the rate of correct nlatching is higher than 98%.