The Research On Real Time Detection Key Technologies For Large Trucks Driving State On Highway

In recent years, major traffic accidents occurred frequently in the freeway, and many accidents caused by large trucks are particularly serious. In order to detect the moving state of large goods vehicles in real time, this paper studied the key technologies of real time detection for large trucks driving state on highway, to provide strong evidence for the handling of traffic accident. In view of the large truck driving test, the paper used the video vehicle detection technology. In view of the large truck driving test, the paper used technology which based on vehicle detection, from three aspects of the detection, tracking and recognition of moving vehicles:1) Vehicle Detection: According to the particularity of highway scenes, this paper proposed the method based on space-time background difference combining with the elimination of the shadow, trace seed compensation, to remove the influence of noise and extract foreground objects clearer. With this method eliminated the scene interference caused by illumination changes, disturbance, interference and shadow.2) Vehicle Tracking: Duo to expressway vehicle speed, traffic flow is relatively stable, and the direction is relatively simple. The main problem is that the vehicle direction is changed, vehicle tracking might be lost. Based on the comparison of current typical tracking algorithms, the Camshift algorithm and Kalman filter were combined to predict the trajectory of the vehicle in this paper, and ensure the vehicle lane change could be accurately tracked.3) Identification of large trucks: It was easy to distinguish large vehicles and small, medium vehicle, because the license plate color were different. However the large vehicles need to be further subdivided, large trucks can be identified by extracting the characteristic parameters of large vehicles. In this paper firstly we used the license plate color to identify the large vehicles and small, medium vehicles. Then the Meanshift regional segmentation feature extraction algorithm was used to segment the frontal images of large trucks and large buses. According to the feature parameters which were got by the regional boundary line of the vehicle image compared with threshold value to judge the large truck and large bus.Based on the above key technology research, the paper develop the real time detetion subsystem for highway large trucks. The system used the MFC technology, C language and Open CV function library to complete. The system mainly included the front equipment communication control, vehicle detection and identification, LED illegal warning management and illegal data transmission management and other functions. Through a lot of experiments proved that the vehicle detection, tracking and recognition technology could be better applied to real-time detection of large trucks on the highway. And the result could provide strong evidence to support the identification of traffic accident responsibility caused by illegal lane driving for large trucks.