摘要
针对脉冲耦合神经网络无法确定最优分割的问题,提出了一种将脉冲耦合神经网络和类间方差准则相结合的图像分割方法。在每次迭代时将脉冲耦合神经网络点火的神经元对应的像素作为目标,未点火的神经元对应的像素作为背景,计算目标和背景之间的类间方差,取类间方差值最大的分割图像作为最终结果。实验结果表明该方法能获得视觉效果较好的分割结果并具有较强的普适性,对一幅大小为256×256的图像进行分割所需要的时间是0.8秒左右。
Aiming at determining the optimal result of Pulse-Coupled Neural Network for image segmentation, a novel method is proposed which combines the PCNN with between-cluster variance. The fired nerves and the unfired nerves of PCNN corresponding to pixels of image are considered as target and the background respectively. The between-cluster variance between the target and the background is calculated at each process of iteration. The optimal segmentation result is obtained when the maximum value of the between-cluster variance is achieved. Experimental results show that the method can achieve better image segmentation and has a common applicability. The simulation time for segmenting an image with the size of 256 by 256 is about 0.8 second.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2005年第5期93-96,共4页
Opto-Electronic Engineering
基金
航空基础科学基金项目(04I53067)
武器装备预研基金项目
高等学校博士学科点专项科研基金项目(20020699014)资助
关键词
图像分割
脉冲耦合神经网络
类间方差
Image segmentation
Pulse-coupled neural network
Between-cluster variance