摘要
脉冲耦合神经网络(PCNN)已广泛应用于图像分割领域,但其参数众多,计算复杂,迭代次数需人工判别,且对低对比度的图像分割效果不理想。针对这些情况提出了算法改进,首先将PCNN的参数简化为单参数可调,然后将图像对比度作为新的判断因子加入最大类间方差(OTSU)的判别式中,以判断PCNN迭代的停止时间。最后利用PCNN的同步点火机制,设计出一种新的双向PCNN算法,对图像分割结果进行优化,去除孤立的噪点,同时使分割边缘更加清晰。计算机仿真结果表明,该方法具有较好的图像分割效果和较强的适应性。
The pulse coupled neural network(PCNN)is widely used in image segmentation in recent years.However,due to the drawbacks of numerous parameters,complex computation,manual distinguishing for the number of iterations and the mediocre adaptability for low contrast images,an improved algorithm was proposed in this paper.First,PCNN parameters were simplified to single adjustable parameters.Then,the image contrast as a new judge factor was added to the OTSU discriminant algorithm to determine PCNN iteration stop time.Finally PCNN synchronization ignition mechanism was applied to design a novel bidirectional PCNN algorithm which can remove isolated noise,and make the edge of the segmentation more clear.Simulation results show that this method has better segmentation result and strong adaptability.
出处
《中国科技论文》
CAS
北大核心
2016年第2期236-240,共5页
China Sciencepaper
基金
国家自然科学基金资助项目(61471311)
关键词
图像分割
双向脉冲耦合神经网络
最大类间方差
红外图像
阈值
image segmentation
bidirectional pulse coupled neural network(BPCNN)
OTSU
infrared image
threshold
