Noisy colored point set matching

We propose a process for determining approximated matches, in terms of the bottleneck   distance, under color preserving rigid motions, between two colored point sets A,B∈R2A,B∈R2, |A|≤|B||A|≤|B|. We solve the matching problem by generating all representative motions that bring AA close to a subset B′B′ of set BB and then using a graph matching algorithm. We also present an approximate matching algorithm with improved computational time. In order to get better running times for both algorithms we present a lossless filtering preprocessing step. By using it, we determine some candidate zones   which are regions that contain a subset SS of BB such that AA may match one or more subsets B′B′ of SS. Then, we solve the matching problem between AA and every candidate zone. Experimental results using both synthetic and real data are reported to prove the effectiveness of the proposed approach.