Scatter search for point pattern matching: A comparative study

Matching of point patterns is a fundamental process prior to many applications such as image alignment, object recognition, and pattern retrieval. When two images are aligned, people prefer to deal with sets of local features instead of pixel arrays to increase the accuracy and save the computational time. Given two point patterns, the aim of point pattern matching (PPM) problem is to find an optimal affine transformation which transforms one point pattern by reference to the other such that a dissimilarity measure between them is minimized. This paper investigates the strengths and weaknesses of applying scatter search to cope with the PPM problem. The performance of the proposed algorithm is evaluated by competing with existing algorithms on synthetic datasets. The experimental results manifest that the proposed algorithm is superior and malleable against varying scenarios.