Research on Binocular Ranging Based on Visible and Infrared Cameras Under Low-Light Conditions

In order to address the challenge of detecting targets and measuring their distance under low-light conditions such as urban night lighting and nighttime road lighting, this paper innovatively proposes a distance measurement system based on visible and infrared binocular imaging. First, pedestrian targets in low-light visible and infrared images are separately detected using YOLOv8. Second, the detected targets are matched through a coarse registration method. Finally, the distances between the system and the matched targets in the original visible and infrared images are calculated using an unequal focal length binocular camera method, based on the calibration of a binocular camera at distances of 2 to 5 meters. The experimental results show that within the range of 2–5 meters, the relative error of binocular image ranging is within 10%. This distance measurement system can effectively measure the distance of matched targets in low-light visible and infrared binocular images. This system has a certain promoting effect on applications such as assisted driving or unmanned driving.