Infrared and visible image fusion method based on LatLRR and ICA

Abstract

To solve the problem of missing lots of texture details in the fusion image, we propose a new fusion method of infrared and visible images based on latent low-rank representation(LatLRR) and independent component analysis(ICA) in this paper. Firstly, the source image is decomposed into low-rank components, sparse components, and noise components by LatLRR. Secondly, ICA is utilized for the low-rank part of infrared image and visible image to obtain the main difference between two source images. Then, the image containing more information is determined by comparing the entropy of two source images and it is employed as a benchmark. Finally, the fused image is accomplished by connecting the benchmark result, the low-rank components, and the sparse components of another image according to the result obtained by ICA. Compared with other fusion methods, experimental results demonstrate that the proposed method has better visual effects and evaluation indicators.