Active Contour Model With Fish Swarm Algorithm

Active contour models have been widely used in image processing and computer vision. By definition of the different active contour models can be divided into two categories: parameter active contour model, geometric active contour model. Parametric active contour models use explicit parameter curves, and geometric active contour model is represented by a high one-dimensional level set implicit function.Convergence speed of parametric active contour is slow; segmentation results are affected by the initial contour in capturing small areas (such as concave and narrow is not recognized), and so easy to fall into local minimum. Fish swarm algorithm does not depend on the parameters of the objective function and does not depend on the initial value. Fish swarm algorithm is parallelism and global optimization. When fish swarm active contour model algorithm is applied to the parameters active contour model, the parameters active contour model will be improved.From the above idea, proposed fish swarm active contour model. Specific tasks in this text as follows:(1) Introduces the image segmentation, the classification of active contour model and development process of active contour model. The development process and he present status of the fish swarm algorithm is introduced.(2) Introduces the basic principles of the fish swarm algorithm and the algorithm flow.(3) Introduces the basic principles of active contour models, mathematical models of active contour model, and focuses on analysis of the external gradient vector flow (GVF).(4) In the fourth chapter, the proposed active contour based on gradient vector flow fish special food domain. The food definition is different from the food definition in the traditional fish swarm algorithm, which uses the objective function value as the food concentration, and which is relative to gray scale of each control point and the gradient vector field of the contour. The improvement food concentration can overcome the difficulty that the weight of the individual fish changes blindly, and can improve the speed of execution. Based on the improving food concentration, the total energy of the individual fish is designed to estimate the effect of the active contour model fish. Then, design and implement the unique behaviors of the individual active contour fish, such as preying behavior, clustering behavior, following behavior, leaping behavior.(5) In the design and implement of the active contour model with fish swarm algorithm, the bulletin board involving in the algorithm has been improved. Two new variables are included in the bulletin board, in order to prevent the best fish leaping and decide when to stop the iteration of the behaviors. By Improving the bulletin board, the algorithm's computational speed improves.(6)The improved fish algorithm contains the revising of the best fish so as to locate the target contour accurately.(7) Using mathematical methods (Markov matrix) show that the algorithm is global convergence and parallelism.(8) Compare the results of the active contour model with fish swarm algorithm, the results of the traditional parameters active contour model (the parameter active contour model basing on variation calculus) and the results of GVF active contour model. And it shows the advancements of the active contour model with fish swarm algorithm.