基于改进PCNN和互信息熵的自动图像分割_魏伟一.pdfVIP

第36 卷 第 13 期 计 算 机 工 程 2010 年 7 月 Vol.36 No.13 Computer Engineering July 2010 ·图形图像处理· 文章编号：1000—3428(2010)13—0199—02 文献标识码：A 中图分类号：TP391 基于改进PCNN 和互信息熵的自动图像分割 1,2 1 魏伟一 ，李战明 (1. 兰州理工大学电气工程与信息工程学院，兰州 730050 ；2. 西北师范大学数学与信息科学学院，兰州 730070) 摘 要：脉冲耦合神经网络(PCNN) 由于其良好的脉冲传播特性在图像分割中得到了广泛应用。针对其需要人机交互通过实验确定其相关参 数等问题，改进PCNN 模型，以像素对比度作为链接矩阵，以互信息作为迭代终止的判决依据，提出基于改进脉冲耦合神经网络的自动图 像分割。实验结果表明，该方法实时性好、自适应性强，分割出的目标轮廓清楚。 关键词：脉冲耦合神经网络；图像分割；图像互信息熵 Automated Image Segmentation Based on Modified PCNN and Mutual Information Entropy WEI Wei-yi1,2, LI Zhan-ming1 (1. College of Electrical Engineering and Information Engineering, Lanzhou Univeristy of Technology, Lanzhou 730050; 2. College of Mathematics and Information Science, Northwest Normal University, Lanzhou 730070) 【Abstract 】For its good property of pulse burst, Pulse Coupled Neural Network(PCNN) is widely used in image segmentation. However, there are such problems in the method as its parameters are decided by experiment, so use the contrast of pixels as model’s link matrix, and use image mutual information entropy as the criterion to terminate iteration to modify standard model. This paper proposes an automated image segmentation based on modified PCNN. Experimental results show that the method is adaptive, its real time ability is good, and target contour is more clear. 【Key words 】Pulse Coupled Neural Network(PCNN); image segmentation; image mut