SimpleFusion: A Simple Fusion Framework for Infrared and Visible Images

• DOI: 10.48550/arxiv.2406.19055
• Publication Date: 2024-06-27
• Authors: Ming Chen; Yuxuan Cheng; Xinwei He; Xinyue Wang; Yan Aze; Jinhai Xiang
• Keywords: FOS: Computer and information sciences; Computer Vision and Pattern Recognition (cs.CV); Computer Science - Computer Vision and Pattern Recognition

Integrating visible and infrared images into one high-quality image, also known as image fusion, is a challenging yet critical task for many downstream vision tasks. Most existing works utilize pretrained deep neural networks or design sophisticated frameworks with strong priors this task, which may be unsuitable lack flexibility. This paper presents SimpleFusion, simple effective framework fusion. Our follows the decompose-and-fusion paradigm, where are decomposed reflectance illumination components via Retinex theory followed by fusion of these corresponding elements. The whole designed two plain convolutional without downsampling, can perform decomposition efficiently. Moreover, we introduce loss detail-to-semantic to preserve complementary information between modalities We conduct extensive experiments on benchmarks, verifying superiority our method over previous state-of-the-arts. Code available at \href{https://github.com/hxwxss/SimpleFusion-A-Simple-Fusion-Framework-for-Infrared-and-Visible-Images}{https://github.com/hxwxss/SimpleFusion-A-Simple-Fusion-Framework-for-Infrared-and-Visible-Images}