基于非抽样Shearlet变换的红外与可见光图像融合方法  

Infrared and Visible Light Images Fusion Algorithm Based on Non-subsampled Shearlet Transform

作　　者：高国荣[1] 刘艳萍[1] 

机构地区：[1]西北农林科技大学理学院

出　　处：《农业机械学报》2014年第3期268-274,共7页Transactions of the Chinese Society for Agricultural Machinery

基　　金：西北农林科技大学人才基金资助项目

摘　　要：针对同一场景红外网像与町见光图像的融合问题，提出了一种基于非抽样Shearlet变换（NSST）的融合算法。首先对源图像进行多尺度、多方向NSST分解，得到低频子带系数和各带通方向子带系数；然后，在局部区域结构相似度的基础上，采用基于局部区域能量的方法选择融合图像的低频子带系数；基于脉冲耦合神经网络（PCNN）对带通方向子带空间频率（sF）的响应而得到的点火次数选择融合图像的带通方向子带系数，得到融合图像的NSST系数；最后经过非抽样Shearlet逆变换得到融合图像。实验结果表明：与其他5种相关的融合方法相比，该方法可获得具有更好视觉效果和更优量化指标的融合图像。Focusing on the fusion problem of infrared and visible light images in the same scene, a novel muhirsensor image fusion algorithm based on the non-subsampled Shearlet transform was proposed. Firstly, the NSST was performed on the source images at different scales and directions, thus the low frequency subband coefficients and varieties of directional bandpass subband coefficients were obtained. Secondly, the low frequency subband coefficients of the fused image were selected based on the local structural similarity and local energy of the two source images, and the bandpass subband coefficients of the fused image were selected based on the firing times of the pulse coupled neural network（PCNN） , so the NSST coefficients of fused image was got. Finally, the fused image was obtained by performing the inverse NSST on the combined coefficients. Quantitative and qualitative analysis of the experimental results demonstrated that the proposed method performs significantly better than the other five related methods.

关 键 词：红外图像 可见光图像 剪切波变换 融合结构相似度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]