摘要
为有效地对受噪声影响的图像进行分割,引入了一种基于改进型遗传算法的局部动态阈值选取算法。该算法以灰度图的最大类间差作为适应度评价函数,将图像分割问题转化成一个最优化问题,利用遗传算法的全局优化的特点,在整个求解空间中搜索到最优分割图像的阈值。针对基本遗传算法的一些不足之处,利用小生境的思想,对其进行了一些优化。实验结果表明,基于小生境技术遗传算法的局部动态阈值选取方法具有较强的鲁棒性。
To effectively accomplish image segmentation tasks for noisy image, one improved algorithm for the selection of the local dynamic threshold based on the niched genetic algorithm was proposed. With the maximum variance method used as the fitness evaluation function, the image segmentation problem was transferred into an optimization issue. By applying the characteristics of the integral optimization of the genetic algorithm, the optimal threshold was searched from the segmentation parameter space. Some optimizations were made to some problems of the basic genetic algorithm by the niche idea. The results of experiment show that the selection of the local dynamic threshold based on the niched genetic algorithm has better robustness.
出处
《武汉理工大学学报（信息与管理工程版）》
CAS
2008年第2期218-221,共4页
Journal of Wuhan University of Technology：Information & Management Engineering
