2026 年 65 巻 1 号 p. 39-45
Traffic monitoring cameras are installed on expressways to check road conditions and respond quickly to traffic accidents. However, the images captured by these cameras are currently monitored by humans, which causes a delay in response due to missed objects. To improve this situation, the automatic detection of road areas and objects on the areas is necessary day and night. Given the above, we propose a method of automatically detecting road areas in nighttime video images. The white line segments that appear when illuminated by the headlights of passing vehicles are detected from the video images, and the entire road area is automatically detected by accumulating information over time. We developed a prototype program based on the proposed method and conducted experiments using real video images to confirm the effectiveness of the method.