Curvelet域自适应脉冲耦合神经网络的图像融合方法  

Method for image fusion based on adaptive pulse coupled neural network in curvelet domain

作　　者：王昊鹏[1,2] 刘泽乾[2] 方兴[3] 张会勇[3] 王晶横[3] 

机构地区：[1]海军航空工程学院武器系统与运用工程系,山东烟台264000 [2]空军航空大学作战指挥系,吉林长春130022 [3]空军航空大学基础训练基地,吉林长春130022

出　　处：《光电子．激光》2016年第4期429-436,共8页Journal of Optoelectronics·Laser

基　　金：军队重点科研(KJ2011215)资助项目

摘　　要：为解决传统图像融合准则不能充分利用图像全局特征的问题,将脉冲耦合神经网络(PCNN)模型用于Curvelet变换的图像融合中,提出了由表征子带图像局部特征的支撑值(SPV)作为刺激PCNN模型的外部激励输入,同时考虑Curvelet变换后低频子带信息与高频子带信息间的相关性,设定PCNN模型参数(连接强度和连接范围)随低频子带图像的特征自适应地变化,并且利用PCNN模型中各神经元的首次点火时间构造融合准则中的显著性度量。用PCNN模型模拟人眼视觉神经系统的生物特性,并利用其全局耦合特性对源图像进行智能地分析判断和融合处理,从而提高融合图像的整体效果。实验结果表明,由于PCNN具有全局耦合特性和脉冲同步特性,因此当它用来参与选取细节系数时,能够更好地利用子带图像的全局信息。In order to solve the problem that the rules of traditional image fusion cannot make full use of global image features,this article used pulse coupled neural network（PCNN）model into the Curvelet transform image fusion,made the Support Value（SPV）of the characterization sub band image characteristics as the external excitation input to stimulate PCNN model,at the same time,after considering the correlation of information between low and high frequency sub band of the Curvelet transform,set PCNN model parameters（strength and range of the connection）to change with the characteristics of the low frequency sub band images adaptively,and using the first ignition timing of each neuron in the PCNN model to structure the significant measure of fusion rule.Using the PCNN model to simulate the biological characteristics of human visual neural system,and use the global coupling features to analyze and process the fusion of the source image intelligently,so as to improve the overall effect of the fusion images.The experimental results show that,when PCNN is used to participate in selecting detail coefficients,due to the global coupling characteristics and pulse synchronization feature,it can use the global information of sub band image better.

关 键 词：图像融合 CURVELET 脉冲耦合神经网络(PCNN) 支撑值(SPV) 自适应 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]