Based On Swarm Intelligence Algorithm For Image Segmentation Method

Now in area of studying on intelligent automatic technique the algorithms which possessed colony and intelligence have come in for more and more attention. As a representative realization of colony intelligence and, particle swarm optimization and ant colony algorithm has received extensive attention from academe. At the end of 1990s ant colony algorithm (ACA) is proposed by Italian scholars M.Dorigo, who is inspired by the ant finding the shortest road from the nest to the destination. As a novel simulated evolutionary algorithm, ant colony algorithm has many strong points as parallel essence, positive feedback and coordination. It has favorable adaptability in solving complex optimization problems and has great development potential. The main characteristics of PSO are simple theory, few parameters, fast convergence and few related knowledge needed. The main drawbacks of PSO are that it is very easy to relapse into local best and premature convergence. PSO has been well applied in many aeras, such as function optimization, neural network training, combinatorial optimization, robot path planning and so on. However, the theory research and application of PSO are not very mature, and there are still many problems to be solved.Image segmentation is a technology and process of partitioning the image into several areas with the similar characteristics and finding the interested object. It is the precondition of image analysis and pattern recognition, and it is the classic problem of image processing. The maximum entropy threshold is widely applied into image segmentation. However the problem of time-consuming computation won't meet real-time requirement when we try to searh optimum multilevel thresholding with the traditional method. The active contour model is a relatively new image segmentation algorithm. The time-consuming and it can not converge to the image concave are the main problem when we compute it with the traditional method. PSO simulated evolutionary algorithm, and the parallel, positive feedback and robust make the PSO suitable for image segmentation.This dissertation research some improved PSO algorithms and ACT, and its application to image segmentation. The main works of the dissertation can be organized as follows:1. Firstly, researching on ant colony algorithm and particle swarm optimization, summarize the studying headway. sum up the applying area of ACA and PSO and the disadvantage. Secondly, then deeply introduce the principle of image segmentation.2. Introduce relative knowledge on clustering. Including summary of clustering, mathematic model, and idea of clustering based on ant's searching for food and the principle based on ant stacks.3. According to clustering and discreteness of ant colony algorithm, a clustering methods based on ant colony algorithm are applying in image segmentation, which origin from the principle of ant stack's setting. Experimental results show the algorithm proposed in this thesis have effective results, and the procedure is simple.4. Based on the theory of information entropy, we proposed a particle swarm optimization algorithm based on information entropy. Validate the availability of algorithms with statndard testing image.