摘要
图像分割是图像处理和图像分析的关键步骤,在图像工程中占据重要地位。模糊C均值聚类(FCM)算法是一种经典的模糊聚类分析方法,但其算法初始聚类原型是随机选取的,从而造成算法性能强烈地依赖聚类原型的初始化,将遗传算法强大的通用性应用于模糊聚类算法,对模糊聚类中心进行编码,然后依据FCM算法的目标函数建立适应度函数,选择适当的交叉率和变异率,最终实现基于模糊聚类遗传算法的图像分割。采用这种方法一方面能较好地解决模糊聚类对初始化敏感的问题,又能在一定程度上提高了分割速度。实验结果表明,该算法具有良好的分割效果。
Image segmentation is a key technology both in image processing and image analysis, which plays an important role in image project. Fuzzy C-means algorithm is one of the most popular methods of clustering analysis. However, the traditional FCM algorithm does not work well because its initial clustering central collection is the stochastic selection. The genetic algorithm which has a powerful universality is introduced. Firstly the fuzzy cluster center is coded,then the fitness function is established according to the object function in FCM algorithm,and under the appropriate crossover rate and mutation rate, the image segmentation based on the genetic fuzzy clustering algorithm is realized. Using the new method,limitation of initial sensitivity has been overcome about fuzzy clustering, and the segmentation speed has been improved to some extent. The experiments show that this segmentation algorithm has achieved a good effect.
出处
《现代电子技术》
2009年第16期120-122,共3页
Modern Electronics Technique
关键词
模糊聚类
遗传算法
图像工程
图像分割
fuzzy clustering
genetic algorithm
image engineering
image segmentation
