一种结合遗传算法的图像分割方法

A Image Segmentation Method Combimng Genetic Algorithm

模糊聚类算法是一种解决图像分割的常见算法,Stelios在模糊C均值聚类算法的基础上提出了FLICM算法,极大地改进了图像分割的效果。基于此,经过分析证明FLICM算法存在的不收敛问题,在此基础上改进了FLICM算法,并提出了结合遗传算法来解决因目标函数复杂度高而无法给出"闭合"迭代公式的问题。从结果来看,该算法不仅克服了FLICM算法不收敛的问题,而且取得了更好的图像分割效果,使得图像细节得到更充分的保留。

Fuzzy c-means clustering algorithm is a common method for solving image segmentation, stelios proposed FLICM algorithm based on fuzzy c-means clustering algorithm, which greatly improved the effect of image segmentation. Based on this, the analysis reveals that the object function of flicm is not conver-gent, so this paper proposed a method that use genetic algorithm to solve the problem, which has a com-plex objective function that cannot obtain the convergent iterative formulas. Experimental results on synthet-ic real- world images showed that this method was not only convergent but also more efficient at providing robustness to noisy image and keeping details.： Fuzzy c-means clustering algorithm is a common method for solving image segmentation, stelios proposed FLICM algorithm based on fuzzy c-means clustering algorithm, which greatly improved the effect of image segmentation. Based on this, the analysis reveals that the object function of flicm is not conver-gent, so this paper proposed a method that use genetic algorithm to solve the problem, which has a com-plex objective function that cannot obtain the convergent iterative formulas. Experimental results on synthet-ic real- world images showed that this method was not only convergent but also more efficient at providing robustness to noisy image and keeping details.： Fuzzy c-means clustering algorithm is a common method for solving image segmentation, stelios proposed FLICM algorithm based on fuzzy c-means clustering algorithm, which greatly improved the effect of image segmentation. Based on this, the analysis reveals that the object function of flicm is not conver-gent, so this paper proposed a method that use genetic algorithm to solve the problem, which has a com-plex objective function that cannot obtain the convergent iterative formulas. Experimental results on synthet-ic real- world images showed that this method was not only convergent but also more efficient at providing robustness to noisy image and keeping details.