基于HMSD与改进PCNN的红外与可见光图像融合

打开文本图片集

关键词：图像融合；红外与可见光；混合多尺度分解；快速交替引导滤波器；脉冲耦合神经网络中图分类号：TP394.1;TH691.9 文献标识码：Adoi：10.37188/OPE.20253309.1481 CSTR：32169.14.OPE.20253309.1481

Infrared and visible image fusion based on HMSD and improved PCNN

REN Pengbai1*，LEI Huiyun 2 ， DANG Jianwu 1，2 ， WANG Yangping²，LIU Qiming 3 ， YANG Li² （1.National Virtual Simulation Experiment Teaching Center for Rail Transit Information and Control， Lanzhou Jiaotong University， Lanzhou 73oo7O，China; 2. School of Electronic and Information Engineering， Lanzhou Jiaotong University， Lanzhou 730070，China; 3. Gansu Xinwangtong Technology Information Co.，Ltd，Lanzhou 73Oo7O， China） * Corresponding author， E-mail： renpb@mail. lzjtu. cn

Abstract：To address the issue of edge and detail information degradation in infrared and visible image fusion caused by limitations such as information loss and data redundancy，a novel approach is presented. Traditional multi-scale domain fusion methods often result in the loss of edge information in both infrared and visible images.This study proposes a hybrid multi-scale decomposition model （HMSD） integrated with an enhanced pulse-coupled neural network （PCNN） for infrared and visible image fusion. The HMSD model，developed by combining the characteristics of fast alternating guided filtering（FAGF）and Gaussian filtering （GF），decomposes the source images into a base layer and three feature maps，each capturing both fine and coarse structures. The fusion of the base layers is performed using a nuclear norm minimization（NNM） fusion rule，while the fusion of the feature maps employs the improved PCNN and regional energy-based rules. Experimental results demonstrate that the proposed method achieves average improvements of 47.6% ， 5.2% ， 6.4% ， 9.4% ， 5.3% ，and 27.3% across spatial frequency， average gradient，correlation coeficient，information entropy，and standard deviation metrics，respectively. This method not only preserves the edge and texture information of the source images but also significantly enhances the visual quality of the fused images..

Key words： image fusion； infrared and visible；mixed multi-scale decomposition； fast alternating guidedfiltering； pulse coupled neural network

1引言

由于技术条件的约束以及拍摄环境的影响，单一设备拍摄难以实现对整个场景的全局性、完整性描述，因此，图像融合技术应运而生[1]。（剩余15369字）