基于PSO和M-S模型的图像分割方法

为得到快速、准确的图像分割方法，提出了一种基于微粒群算法（PSO）的主动轮廓线模型和Mumford-Shah（M-S）模型的方法。利用PSO方法对主动轮廓线模型的蛇点寻优，使其快速收敛到图像边缘附近，得到目标的粗糙轮廓，作为M-S模型的初始水平集；并将窄带方法引入M-S模型的计算，快速得到准确的分割结果。该方法克服了主动轮廓线模型对初始曲线敏感、不能收敛到物体的凹陷边缘、对噪声敏感问题和M-S模型需要对所有图像数据进行计算且计算量大等问题。实验结果表明了该方法的可行性和有效性。

Image Segmentation Approach Based on PSO and Mumford-Shah Model

An image segmentation approach is proposed for gaining rapid and accurate image segmentation.Particle swarm optimization(PSO) gets the optima of snake points that rapidly converge near image edge.Rough contour of object is received and used for initial zero level set of Mumford-Shah(M-S) model.To get accurat result quickly,the narrow band method is used in M-S model.The problems of traditional active contour models,such as the sensitivity to initial position and noise,and the poor convergence to concavities,are solved.In each step of iteration,only the data in a narrow band is dealed with,so the computational complexity is reduced.Experimental result shows the effectiveness and feasibility of the method.