Computer Engineering and Applications ›› 2009, Vol. 45 ›› Issue (35): 202-204.DOI: 10.3778/j.issn.1002-8331.2009.35.061

• 图形、图像、模式识别 • Previous Articles     Next Articles

Image segmentation using PCNN and maximal correlative criterion

NIE Ren-can1,LI Sha2,NIE Cai-ren3   

  1. 1.Department of Communication Engineering,Information College,Yunnan University,Kunming 650091,China
    2.Department of Equipment,Chengdu Military Area,Chengdu 610031,China
    3.Yunnan Normal University,Kunming 650222,China
  • Received:2008-07-09 Revised:2008-10-17 Online:2009-12-11 Published:2009-12-11
  • Contact: NIE Ren-can

脉冲耦合神经网络与最大相关准则的图像分割

聂仁灿1,李 莎2,聂彩仁3   

  1. 1.云南大学 信息学院 通信工程系,昆明 650091
    2.成都军区装备部,成都 610031
    3.云南师范大学,昆明 650222
  • 通讯作者: 聂仁灿

Abstract: Based on the PCNN(Pulse Coupled Neutral Network) model,propose an image pre-processing method based on Canny algorithm to enhance the blurry edge for image,and strengthen or restrain the capacity of capturing each other for a similar cluster of corresponding neurons.Using the maximal correlative criterion to compute the attenuation threshold,control the capturing intensity each other for neurons,and avoid the excessive smoothing aroused by through capturing for segmentation image,a new image segmentation method is put forward successfully.A good image segmentation result,which shows more image segmentation details and is not influenced easily by neuron parameters compared with other corrective references,is gained using the method.

摘要: 基于脉冲耦合神经网络(Pulse Coupled Neutral Network,PCNN)模型,提出了一种基于Canny算法的图像预处理方法,增强了图像的模糊边缘,提高或抑制了对应神经元的相似群捕获能力,利用最大相关准则确定神经元的衰减阈值和实现神经元相似群捕获强度的控制,避免了神经元完全捕获对分割图像的过平滑,成功实现了灰度图像的自动分割。该方法得到的分割图像取得了较好的结果,体现了更多细节,与相关文献相比,大大减少了神经元参数对分割结果的影响。

CLC Number: