@article{liu2023coconet,
title={Coconet: Coupled contrastive learning network with multi-level feature ensemble for multi-modality image fusion},
author={Liu, Jinyuan and Lin, Runjia and Wu, Guanyao and Liu, Risheng and Luo, Zhongxuan and Fan, Xin},
journal={International Journal of Computer Vision},
pages={1–28},
year={2023},
publisher={Springer}
}
论文级别:SCI A2
影响因子:19.5
[论文下载地址]
作者提出了一种耦合对比学习网络CoCoNet,这是一个【通用】的图像融合网络。
使用耦合对比学习来指导模型区分目标以及纹理细节,并且采用了一种测量机制来计算源图像的比例重要性,以生成数据驱动的权重并应用于损失函数之中。
image fusion, infrared and visible image, unsupervised learning, contrastive learning
图像融合,红外和可见光图像,无监督学习,对比学习
保持互补信息,消除冗余信息。
使用数据驱动机制计算信息保留度,以提高融合结果和源图像强度和细节的一致性。
使用多级注意力模块(multi-level attention module ,MAM)避免融合过程中的特征退化。
参考链接
[什么是图像融合?(一看就通,通俗易懂)]
作者的思路是,将红外图像的显著目标作为正样本,将可见光图像的显著目标作为负样本;同理,将可见光图像的的背景作为正样本,红外图像背景作为负样本。
基于TNO数据集人工标注掩膜,设 M \mathcal M M为前景的显著掩膜, M ˉ \bar {\mathcal M} Mˉ为背景的显著掩膜。
红外图像×前景掩膜可以得到显著目标,可见光图像×背景掩膜得到了背景信息。
作者选用预训练VGG-19代表G,将此处的损失函数定义为:
N和M分别为每个正样本的VGG层数和负样本数。
μ i \mu_i μi代表融合图像的前景特征 G i ( I F ⊙ M ) G_i(I_F\odot \mathcal M) Gi(IF⊙M)
μ i + \mu_i^+ μi+和 μ i m − \mu_i^{m-} μim−分别是正样本和负样本, μ i + = G i ( I R ⊙ M ) \mu_i^+=G_i(I_R\odot \mathcal M) μi+=Gi(IR⊙M) μ i m − = G i ( I V m ⊙ M ) \mu_i^{m-}=G_i(I_V^m \odot \mathcal M) μim−=Gi(IVm⊙M)
m m m代表第m个负样本, ∣ ∣ ⋅ ∣ ∣ 1 ||·||_1 ∣∣⋅∣∣1是L1范数。
同理,在背景部分,将可见光图像背景作为正样本,红外图像背景作为负样本。细节约束的目标函数为:
v i v_i vi代表融合图像的背景特征 G i ( I F ⊙ M ˉ ) G_i(I_F\odot \bar {\mathcal M}) Gi(IF⊙Mˉ)
v i + v_i^+ vi+和 v i m − v_i^{m-} vim−分别是正样本和负样本, v i + = G i ( I V m ⊙ M ˉ ) v_i^+=G_i(I_V^m \odot \bar {\mathcal M}) vi+=Gi(IVm⊙Mˉ) v i m − = G i ( I R ⊙ M ˉ ) v_i^{m-}=G_i(I_R \odot \bar {\mathcal M}) vim−=Gi(IR⊙Mˉ)
自适应损失函数=结构损失+强度损失
以往的方法,权重参数都是手工设计的经验值,本文作者设计了一种考虑数据特性的自适应损失。
一方面,为了保留纹理细节,采用平均梯度法(AG)优化SSIM损失的权重参数 σ \sigma σ
∇ h I F {\nabla _h}{I_F} ∇hIF和 ∇ v I F {\nabla v}{I_F} ∇vIF分别代表融合图像从水平方向和垂直方向的一阶微分(梯度)
另一方面,采用图像熵(EN)更新强度损失:
L L L代表给定图像的灰度值, p x p_x px像素处于对应灰度值的概率。
EN是像素级计算的图像信息量,与MSE约束密切相关,因为MSE约束也是像素级的约束。
综上,损失函数为
L i r \mathcal L_{ir} Lir和 L v i s \mathcal L_{vis} Lvis是两对对比损耗。
图像融合数据集链接
[图像融合常用数据集整理]
参考资料
[图像融合定量指标分析]
✨✨✨参考资料
✨✨✨强烈推荐必看博客[图像融合论文baseline及其网络模型]✨✨✨
更多实验结果及分析可以查看原文:
[论文下载地址]
[LRRNet: A Novel Representation Learning Guided Fusion Network for Infrared and Visible Images]
[(DeFusion)Fusion from decomposition: A self-supervised decomposition approach for image fusion]
[ReCoNet: Recurrent Correction Network for Fast and Efficient Multi-modality Image Fusion]
[RFN-Nest: An end-to-end resid- ual fusion network for infrared and visible images]
[SwinFuse: A Residual Swin Transformer Fusion Network for Infrared and Visible Images]
[SwinFusion: Cross-domain Long-range Learning for General Image Fusion via Swin Transformer]
[(MFEIF)Learning a Deep Multi-Scale Feature Ensemble and an Edge-Attention Guidance for Image Fusion]
[DenseFuse: A fusion approach to infrared and visible images]
[DeepFuse: A Deep Unsupervised Approach for Exposure Fusion with Extreme Exposure Image Pair]
[GANMcC: A Generative Adversarial Network With Multiclassification Constraints for IVIF]
[DIDFuse: Deep Image Decomposition for Infrared and Visible Image Fusion]
[IFCNN: A general image fusion framework based on convolutional neural network]
[(PMGI) Rethinking the image fusion: A fast unified image fusion network based on proportional maintenance of gradient and intensity]
[SDNet: A Versatile Squeeze-and-Decomposition Network for Real-Time Image Fusion]
[DDcGAN: A Dual-Discriminator Conditional Generative Adversarial Network for Multi-Resolution Image Fusion]
[FusionGAN: A generative adversarial network for infrared and visible image fusion]
[PIAFusion: A progressive infrared and visible image fusion network based on illumination aw]
[CDDFuse: Correlation-Driven Dual-Branch Feature Decomposition for Multi-Modality Image Fusion]
[U2Fusion: A Unified Unsupervised Image Fusion Network]
综述[Visible and Infrared Image Fusion Using Deep Learning]
[图像融合论文baseline及其网络模型]
[3D目标检测综述:Multi-Modal 3D Object Detection in Autonomous Driving:A Survey]
[CVPR2023、ICCV2023论文题目汇总及词频统计]
✨[图像融合论文及代码整理最全大合集]
✨[图像融合常用数据集整理]
如有疑问可联系:[email protected];
码字不易,【关注,收藏,点赞】一键三连是我持续更新的动力,祝各位早发paper,顺利毕业~