An Improved Quantum Particle Swarm Optimization Algorithm And Its Application On Image Segmentation

In real life, a lot of computational problems can be classified into optimization problems, the vast majority of them are NP-hard. It is difficult to get the global optimal solution in polynomial time for the problems of this class. In the past few decades, in order to solve these problems within an acceptable time people proposed many intelligent optimization algorithms, such as genetic algorithm, ant colony algorithm, simulated annealing algorithm, particle swarm optimization. Among them,the particle swarm optimization algorithm has the characteristics of strong global search capability and simple calculation, so attracts more attention of the researchers.On the other hand, due to the characteristics of quantum parallelism, quantum computation can accelerate the execution speed of many classical algorithms. In recent years, some quantum intelligent optimization algorithms based on quantum computation theory are proposed. Quantum particle swarm optimization algorithm is one of them.Based on the analysis of the classical and quantum particle swarm optimization algorithm, this paper proposes an Improved Quantum Particle Swarm algorithm(IQPSO), and applies it in multiple-threshold image segmentation problem. The related work is as follows:1. We study the basic principle of particle swarm, quantum particle swarm algorithms and some improved versions, analyze the iterative process of these algorithms, then find the drawbacks: for the high-dimensional optimization problems, due to a lack of population diversity, the algorithms are easy to fall into local optimum; In the later stages of the iteration, the global search ability becomes weak, and the search methods have limitations.2. Aiming at the drawbacks, we present two improved strategies: put forward a new method to calculate the point of interest and the characteristic length of the potential well, improve the search strategy to avoid the falling into local optimum; introduce a crossover operation to improve the population diversity and enhance the global search ability.3. the most between-cluster variance method is a most used method to calculate the multiple threshold value of image segmentation. We select it as the evaluation function to determine the threshold values. So this problem can be reduced to optimization problem without constraints. The experimental results show that when we use the IQPSO to solve the problem, it can greatly improve the time efficiency under the premise of good segmentation effect.