一种改进的PCNN图像融合算法  

Improved image fusion algorithm based on PCNN

作　　者：郭传奇[1] 汪文革[2] 储彬彬[3] 

机构地区：[1]解放军炮兵学院四系,合肥230031 [2]解放军炮兵学院二系,合肥230031 [3]解放军汽车管理学院,安徽蚌埠233011

出　　处：《计算机工程与应用》2011年第19期176-178,共3页Computer Engineering and Applications

摘　　要：针对使用小波变换及简单融合规则的图像融合算法的不足,提出了一种改进的基于脉冲耦合神经网络(Pulse Coupled Neural Networks,PCNN)融合规则的非下采样轮廓波变换(Nonsubsampled Contourlet Transform,NSCT)图像融合方法。对已配准待融合图像进行NSCT分解,采用改进的PCNN融合规则对Contourlet域系数进行融合,得到融合图像的NSCT系数,经逆变换重构得到融合图像。实验结果表明该算法在主观视觉和客观评价指标上都取得了较好的融合效果。To overcome the defects of image fusion algorithm using Wavelet transform and simple fusion rules,an improved image fusion algorithm based on pulse coupled neural networks and non-subsampled Contourlet transform is put forward in this paper.The input registered images are decomposed by NSCT.The Contourlet domain coefficients in different scales and directions are fused by improved fusion rules based on PCNN.The fusion image is acquired by recomposing the fused coefficients.Experiments show that this scheme has acquired promising results both in the subjective vision aspect and the objective assessment parameters.

关 键 词：图像融合 多尺度几何分析 非下采样轮廓波变换 脉冲耦合神经网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]