A driver support system to prevent traffic accidents caused by optical illusions

Optical illusions during night driving have a possibility to induce serious traffic accidents. This study, therefore, aims to build an automatic assistance system to detect the visual illusions and to alert drivers to potential dangers. In simulation studies, the proposed support system was able to decide the correct positional relationship between two automobiles from tail lamps and silhouette images. The Speeded-Up Robust Features (SURF) technique, which can identify image features, was also applied to actual driving scenes with external noises (e.g., streetlights, traffic signs, etc.). In future works, machine learning based on the image features could realize a robust assistance system.