Heuristic Optimization Algorithms for a Tree-Based Image Dissimilarity Measure

In this paper, we present an application of three heuristic optimization algorithms to computing tree-based image dissimilarity. Genetic algorithm, particle swarm optimization and simulated annealing have been applied to optimize a black-box function which aims to determine a difference between two trees, constructed upon binary images. Presented results show that the particle swarm optimization achieved the best results. Both PSO and the simulated annealing outperformed the genetic algorithm. We also draw conclusions on parameter adjustment for the considered methods.