Study On Vehicles Detection Technology

With the accelerating urbanization and the rising number of vehicles, traffic congestion has become a major problem faced by cities worldwide. Energy consumption and air pollution caused by traffic congestion has become the main barrier of restricting the city economic and social development. It is an effective method to develop intelligent transportation system. Acquiring the vehicles information on rode is an important link of ITS. With the development of computer and image processing technology, vehicles detecting method based on video has been paid more attention to and has become the important way to acquiring the information of roads. This paper researches the vehicles detecting technology based on video. The main research contents of this paper are as fallows:1. This paper researches the algorithm of background extracting and updating algorithm. First, a vehicle detection approach based on single lane video is approached. In this algorithm, frame difference between tow frames is in stand of that of more frames to Judging the seiendheit of vehicles. This algorithm can effectively deal with the challenges due to lighting changes, shadows, empty holes and breakage in the difference image caused by the vehicle’s uniform color and intensity. Then, to realize extracting and adaptive updating of the background of multilane road and large vision video rapidly, this algorithm is applied in the algorithm based on block method.2. This paper researches the algorithm of vehicles detecting algorithm based on video sequence of single lane road. An effective vehicle detection approach based on improved self-adaptive background subtraction is proposed. This approach can solve the question that it is different to distinguish the vehicles with similar grayscale of the background from the background and this method is excellent in extracting vehicle image frames and detecting the traffic flow under different illumination conditions.3. This paper researches the algorithm of vehicles detecting algorithm based on video sequence of multilane road. To deal with the challenge that vehicles are adhered by shadows in the segmentation result of background subtraction method, an approach based on averaging the results of frame difference of multiple frames is proposed, which can remove most area of the shadow and improve the segmentation effect. Then an algorithm of detecting the traffic flow based on virtual loop is improved, which can improve the detection rate. The test result of different videos in various environments shows that the algorithms proposed and improved by this paper has good effect.