BAIFA: A Brightness Adaptive Image Fusion Algorithm for Robotic Visual Perception

In robot tasks, camera is widely used for scene recognition and localization. However, it is still a challenging problem for robot vision working well in low-brightness environments. We propose a brightness adaptive image fusion algorithm (BAIFA) which fuses one RGB image and one infrared image to improve the quality of image for robotic visual perception. A weight function is presented to calculate the image brightness weight to balance RGB and infrared images in fusion. To verify the proposed algorithm, comparisons with three image fusion algorithms are made in different brightness environments. Experimental results show that the proposed BAIFA is able to effectively preserve image contrast and target contour, which is more robust than others in various brightness environments. Furthermore, a case study shows that, visual perception is improved with our method and the fused image can also provide visual data in mobile robots.