Study on Leading Vehicle Detection at Night Based on Multisensor and Image Enhancement Method

Low visibility is one of the reasons for rear accident at night. In this paper, we propose a method to detect the leading vehicle based on multisensor to decrease rear accidents at night. Then, we use image enhancement algorithm to improve the human vision. First, by millimeter wave radar to get the world coordinate of the preceding vehicles and establish the transformation of the relationship between the world coordinate and image pixels coordinate, we can convert the world coordinates of the radar target to image coordinate in order to form the region of interesting image. And then, by using the image processing method, we can reduce interference from the outside environment. Depending on D-S evidence theory, we can achieve a general value of reliability to test vehicles of interest. The experimental results show that the method can effectively eliminate the influence of illumination condition at night, accurately detect leading vehicles, and determine their location and accurate positioning. In order to improve nighttime driving, the driver shortage vision, reduce rear-end accident. Enhancing nighttime color image by three algorithms, a comparative study and evaluation by three algorithms are presented. The evaluation demonstrates that results after image enhancement satisfy the human visual habits.