Tracking Patterns with Particle Swarm Optimization and Genetic Algorithms

Abstract

The template matching is an important technique used in pattern recognition. The goal is to find a given pattern, of a prescribed model, in a frame sequence. In order to evaluate the similarity of two images, the Pearson's Correlation Coefficient (PCC) is used. This coefficient is calculated for each of the image pixels, which entails an operation that is computationally very expensive. In order to improve the processing time, this paper proposes two implementations for template matching: one using Genetic Algorithms (GA) and the other using Particle Swarm Optimization (PSO) considering two different topologies. The results obtained by the proposed methodologies are compared to those obtained by the exhaustive search in each pixel. The comparison indicates that PSO is up to 236x faster than the brute force exhausted search while GA is only 44x faster, for the same image. Also, PSO based methodology is 5x faster than the one based on GA.