基于模糊熵的改进型PCNN图像分割方法  

A New Method of an Improved PCNN Model for Image Segmentation Based on Fuzzy Entropy

作　　者：赵勇[1] 陈立潮[1] 张英俊[1] 李鑫环[1] 

机构地区：[1]太原科技大学计算机科学与技术学院,山西太原030024

出　　处：《计算机技术与发展》2009年第10期141-144,共4页Computer Technology and Development

摘　　要：针对传统的PCNN在图像分割中需要设定大量的参数,并且分割的最佳迭代次数无法自动确定等问题,简化了传统PCNN模型的接受部分,改进了PCNN的连接部分,改变了PCNN的阈值衰减方式,并利用最大模糊熵作为最佳分割迭代次数的判定准则,提出了一种新的PCNN改进模型,从而实现了PCNN的自动精确分割。对各类图像的实验结果表明,该方法能够自动确定循环迭代次数和自动选取最佳阈值,与基于最大香农熵的PCNN分割方法相比,该方法具有收敛速度快、分割精度高、分割效果好等特点。A new improved PCNN model is proposed to overcome the problem existing in the image segmentation of PCNN. It doesn＇t only simplify the acceptant part of the traditional PCNN and improves on the pontes of PCNN and changes the threshold attenuation mode of PCNN, but also utilizes the maximum fuzzy entropy as the determinant rule of the best segmentation iterations. Therefore, the improved PCNN can implement the segmentation results accurately and automatically. Simulation experiment on various kinds of images indicates that the proposed method can confirm the circulatory iterations and choose the best threshold automatically. Compared to the maximum Shannon entropy, the proposed method has higher convergence speed and segmentation accuracy, and also achieves a better segmented effect.

关 键 词：改进的PCNN 模糊熵 图像分割 阈值 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]