Vehicle Detection Algorithm for Lane Change Decision Aid System

We have developed a Lane Change Decision Aid System (LCDAS), which detects vehicles and motorcycles behind in adjacent lanes with a single camera and informs the driver of dangerous situations during lane change maneuvers. A key technology in LCDAS depends on how it can exactly detect approaching vehicles in spite of weather and environment change. Especially, it's not proper to apply same image analysis algorithm against environment change such as day/night/rainy weather because camera image is very sensitive to illumination condition and weather condition. Therefore it is very important to choose each mode according to vehicle's signals for improving the performance. In this paper, we propose an adaptive LCDAS against weather and environment change. For this purpose, we divide each mode and vehicle signal (wiper, headlamp and vehicle speed sensor) is used as an input and the right/left turn signal is used to choose left/right camera image. The experiments show that the LCDAS can be utilized as a device which assists safety driving by detecting rear side vehicles under weather and environmental change.