Dense Fusion of Infrared and Visible Images

This paper introduce a systematic learning architecture for the fusion of infrared and visible images.. Infrared and visible images are used as input images. Visible image capture reflected light and infrared capture thermal radiation. So the fused image of visible and infrared images provide better information and quality. Architecture have three main sections. The first section is encoding network. Inside the encoder network, conventional layers and dense block are present. The encoder section will extract rough as well as deep features from source images. Second section is Fusion layers. It follows two strategies, and designed to fuse these features. There are two fusion strategies, one is addition strategy and other one is l1-norm strategy is doing. Third section is decoder, it reconstruct the fused images. The final out is more informative and accurate than any single source image, and it consists of all the necessary information. The paper compare the proposed model with the existing fusion technology and also discuss the future scopes of proposed model.