红外与可见光图像融合数据集

这是红外与可见光的数据集，原作者在github有仓库https://github.com/Linfeng-Tang/MSRS我在自己实验的基础上进行了修改删减，然后添加了TNO数据集作为测试集，使用train文件夹进行模型的训练，使用test文件夹下图像进行模型测试。
目前我使用的是对抗生成网络进行图像融合，然后就是基于对抗生成网络进行改进,目前输入图像使用的是单通道的灰度图，使用注意力模块提取通道二者相关性，然后经过auto encoder结构进行融合图像的生成

This is an infrared and visible light dataset. The original author has a GitHub repository at https://github.com/Linfeng-Tang/MSRS. I have modified and pruned this dataset based on my own experimental work, and additionally incorporated the TNO dataset as the test set. The `train` folder is used for model training, while images stored in the `test` folder are utilized for model testing. Currently, I employ generative adversarial networks (GANs) for image fusion tasks, and have developed improvements based on GAN architectures. Single-channel grayscale images are adopted as the input format. An attention module is used to extract the channel-wise correlation between the two input images, and an autoencoder structure is then applied to generate the final fused image.

搜集汇总

背景与挑战

背景概述

该数据集是用于红外与可见光图像融合任务的公共数据集，基于MSRS和TNO数据集构建，包含训练集和测试集，适用于对抗生成网络等深度学习模型的训练与评估。数据集以单通道灰度图形式提供，强调通过注意力模块和auto encoder结构进行图像融合，旨在提升融合图像的质量和相关性提取能力。