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

Particle swarm optimization (PSO) algorithm is inspired by social behavior of bird flocking or fish schooling. It is a population-based, self-adaptive search optimization technique. As a kind of swarm intelligence, it has been proven to be a powerful global optimization method. PSO algorithm has attracted a lot of attention from researchers around the world since its introduction, as its advantages such as rapid convergence towards an optimum, simple computing, easy implementation. Now it has already been successfully applied in many areas, such as image segmentation, function optimization, artificial neural network training, fuzzy system control. However, PSO does exhibits some disadvantages: it sometimes is easy to be trapped in local optima, and the convergence rate decreased considerably in the later period of evolution, thus the global optimal solution can't be achieved.Image segmentation is regarded as an important step in object examination and character recognition. The main objective is to extract objects of interest from an acquired image, so it provides the evidence to the subsequent processing of computer vision. Several methods are proposed from different theoretical point of view for image segmentation. Image threshold segmentation is a powerful tool for image segmentation for its simplicity. However the problem of time-consuming computation won't meet real-time requirement when we try to search optimum multilevel thresholding on a multimodal histogram of a complex image. It's a key for effective image segmentation to exactly determine those thresholding. So it is a difficult problem for us to quickly and exactly search optimum multilevel thresholding for image segmentation. However, to quickly and exactly determine optimum combination of multilevel thresholding, which can segment the image efficiently and meet real-time requirement, we must explore an effective and rapid algorithm to solve the problem of image segmentation based on multilevel thresholding. Based on the former research, the author studies the improvement of particle swarm algorithm and its application on image segmentation. The main works of the dissertation can be organized as follows:Part one introduces in detail the status of particle swarm algorithm and image segmentation respectively, followed by the relative basic concept and the main works of this dissertation.In part two, in order to seek an improved PSO which can greatly increase the convergence velocity of algorithm and increase the global search capability. A novel improved PSO algorithm is presented in this dissertation. The algorithm defines two factors, the evolution speed factor and aggregation degree factor, which determine and change inertia weight of velocity updating formula of the particle dynamically. The convergence velocity of this improved PSO algorithm is enhanced to continue optimizing when the swarm got stagnated for last few iterations by replacing some worst particles in fitness values by the copies of better particles. It is proven that the improved algorithm has high performances in convergence speed and global search capability on different benchmark optimization functions. The presented algorithm is applied to image segmentation based on multilevel thresholding and it is proven to be effective because the experimental result shows the algorithm can exactly determine the optimum multilevel thresholding. It is suitable for a multimodal histogram of a complex image.