摘要
脉冲耦合神经网络(Pulse-coupled neural network,PCNN)可有效地用于图像分割。为获得满意分割效果,PCNN需要选取适当的参数,目前其参数往往通过反复试凑确定。针对这一问题,基于改进的PCNN模型,提出结合图像灰度直方图,以最大交叉熵函数作自适应遗传算法的适应度函数,采用自适应遗传算法搜索最优门限阈值函数的图像分割算法。实验结果表明,该算法可以有效地实现多目标图像分割,且分割效果优于多阈值Ostu算法。
Pulse-coupled neural network（PCNN） can be used for image segmentation. It is necessary to select the appropriate parameters to achieve satisfactory segmentation results when PCNN is applied to image segmentation. Up to now,the parameters of PCNN are always manually adjusted. To solve this problem, a multi-target image segmentation algorithm based on the improved PCNN is proposed. It combines the gray histogram of images, uses the maximal cross-entropy function as the fitness function of adaptive genetic algorithm ,and adopts adaptive genetic algorithm to search the optimal threshold function, Experimental results show that the proposed algorithm can effectively complete multi-target image segmentation, and its segmentation results are superior to that of the multi-threshold Ostu algorithm.
出处
《数据采集与处理》
CSCD
北大核心
2009年第4期536-542,共7页
Journal of Data Acquisition and Processing
关键词
脉冲耦合神经网络
多目标图像分割
自适应遗传算法
灰度直方图
最大交叉熵
pulse-coupled neural network（PCNN）
multi-target image segmentation
adaptive genetic algorithm
gray histogram
maximal cross-entrop
