摘要
目的解决区域生长简化脉冲耦合神经网络(PCNN)算法中由于阈值参数选取不当导致的分割不足与过分割问题。方法在区域生长简化PCNN算法中引入熵来刻画图像的信息量。结果避免了对阈值参数选取。结论基于信息量的PCNN改进算法在分割精度、算法的稳定性等方面均优于简化区域生长PCNN算法。
Objective To solve the threshold parameter selection problems between over-segmentation and less-segmentation in simplified region growing pulse coupled neural network（PCNN）.Methods Image information was introduced to the simplified region growing PCNN described by the concept of entropy.Results The selection of threshold parameter was avoided.Conclusion It is demonstrated that the PCNN modified algorithm based on image information is better than simplified region growing PCNN algorithm in segmentation accuracy and stability of the algorithm
出处
《生物医学工程与临床》
CAS
2010年第6期485-488,共4页
Biomedical Engineering and Clinical Medicine
基金
福建省自然科学基金项目(2008J0312)
南京军区"十一五"计划课题项目(06MA99)
南京军区重点课题(08Z021)
关键词
脉冲耦合神经网络
熵
区域生长简化PCNN模型
图像信息
图像分割
pulse coupled neural network
entropy
simplified region growing PCNN model
image information
imagesegmentation
作者简介
林亚忠（1973-），男，博士，高级工程师，主要从事计算机图像处理、模式识别与数据挖掘研究。
