Information Fusion of Infrared Image and Visible Image for Robot Localization Under Extremely Low Illuminati on

In scenarios characterized by diminished ambient lighting, a solitary source image proves inadequate for precise localization of the robot, thereby potentially impeding subsequent visual SLAM (Simultaneous Localization and Mapping) endeavours. Although infrared cameras offer the advantage of capturing discernible images irrespective of lighting conditions, they may concurrently introduce distortions in the background and finer details of the image. This study introduces a novel approach for enhancing robot localization under low illumination conditions by integrating infrared imagery with low-light information. Our method employs a new weighted fusion strategy and works in conjunction with a photometric sensor to ensure that the robot has sufficient image information in low environments with different low illumination intensities and lighting variations. The efficacy of the proposed system is assessed based on the requisite number of feature points essential for accurate robot localization and is benchmarked against established methodologies. Experimental scrutiny of the weighted fusion technique delineated in this paper corroborates its ability to yield a higher count of feature points conducive to robot localization while concurrently furnishing a more holistic depiction of the image.