Modified strategy to inertia weight in PSO for searching threshold of Otsu rule

In order to searching the best threshold of Otsu rule rapidly, a modified strategy to the inertia weight in Particle Swarm Optimization (PSO) is introduced. The strategy correlates the inertia weight in PSO with the globally best position and the individual best position in particle swarm and modifies the inertial weight by the distance between the best individual position and the globally best position. With the reduction of the distance between the globally best position and the individual best position, the inertia weight of the individual particle will be reduced, vice versa. The experimental results indicate that this algorithm decreases computational work and improves average convergence rate by 21.0726% as compared with that of traditional methods. It can speed up on-line image segmentation using this method.