Visible and Infrared Image Fusion Using Anisotropic Diffusion and Weight Map Construction

摘要

Infrared and visible image fusion could be widely used in military, navigation and medical fields. Infrared image contains thermal radiation information while visible image contains detailed textures of the target and scene. Information elements from these images are complementary. The fusion aims to transfer as much complementary information as possible to fused image. Therefore, the fused image could achieve a better performance than source images in applications such as target detection and camouflage recognition. In this paper, a new method based on anisotropic diffusion and weight map construction is proposed. Considering the characteristic differences between infrared and visible images, different parameters of anisotropic diffusion are used when decomposing those images. Either image is decomposed into base layer and detail layer after anisotropic diffusion. Different rules of weight map construction is used to fuse base layer and detail layer, respectively. Final fused image is generated through linear combination of fused base layer and detail layer. Experiments are conducted to compare proposed method with the traditional and other recently proposed methods according to subjective and objective evaluation criterion. Result shows that the performance of the proposed method is comparable or superior to other methods.