An Improved Fusion Method For Infrared And Low-Light Level Visible Image

In this paper, we propose an improved image fusion algorithm for infrared and visible images acquired under low illumination. Firstly, single scaled Retinex (SSR) algorithm is applied to enhance visible images. And then, both infrared image and the enhanced visible image are decomposed into multilayer images by Laplacian pyramid algorithm. Finally, the decomposed images are fused with the method based on regional features measurement. The experiment results show that even the source images are with low contrast and low illumination, the proposed algorithm can obtain fused images with highly enhanced contrast, which can easily distinguish targets from background, and is better able to preserve texture and detail from source images. Moreover, this SSR-Laplacian algorithm is universal for near-infrared images as well as thermal-infrared images.