dc.description.abstract | In the recent years, the study on safe driving becomes more and more important. To aid the safe driving, we use a camera mounted on a vehicle to capture road scenes for detecting preceding vehicles.
The proposed system consists of three parts. In the first part, we use horizontal edge and underneath shadow of a possible preceding vehicle to locate the vehicle. We then use the vertical edge and symmetry property to detect the vehicle. Finally, we represent the possible vehicle by a detected box. In the second part, we use the lane width, vehicle width, symmetry, and color variance of pixels in the detected box to verify the vehicle. In the third part, we track and search vehicles in a narrow area. The first and third parts are independently executed following by the second part. That is we detect or track a preceding vehicle based on the current situation, then verify the detected or tracked objects.
In the experiments, the proposed detector is evaluated on several different weather conditions such as sunny day, misty day, dusky day, cloudy day, and rainy day. From the experimental results, we find that the proposed approach can robustly and accurately detect or track the preceding vehicles in real time. | en_US |