PSO versus GAs for fast object localization problem

Particle swarm optimization (PSO) and genetic algorithms (GAs) are two kinds of widely used evolutionary compution techniques. In this paper, a particle swarm optimizer is implemented and compared to a genetic algorithm for the object localization problem. The problem of object localization can be formulated into an integer nonlinear optimization problem (INOP). We respectively expand the basic PSO and GA to solve the formulated INOP. Experiments were made on a set of 42 test images with complex backgrounds. The results show that although GA and PSO share many common features, PSO is more suitable for the problem than GA.