Research Of Traffic Incident Detection Algorithm Based On Visual Saliency

As an important part of the Intelligent Traffic System,the Automatic Incidents Detection system plays an important role in avoiding the occurrence of traffic accidents and their and their treatment and control.In recent years,with the application of computer vision and digital image processing technology in traffic incident detection,the traffic incident detection algorithm based video has gradually become a hot topic in the field of intelligent transportation.The key technologies of traffic incident detection algorithm mainly include moving object detection,moving object tracking and object behavior recognition.This paper studies the above key technologies and designs smart traffic incident detection algorithms that meet the requirements of reliability,accuracy,and real-time.In the aspect of moving object detection,this paper proposes a dynamic salient model based on spatiotemporal mixing,which combines the advantages of the static salient model and the dynamic salient model.Static saliency and dynamic saliency were obtained through the Gabor filter and the quaternary Fourier transform to detect the possible occurrence of traffic incidents.In the aspect of moving object tracking,a Kalman filter tracking algorithm based on multifeature matching is proposed in this paper.The Kalman filtering algorithm is used to predict the characteristic value of the object at the next time,then the object is searched and matched,establish the corresponding relation of the object in the continuous frame,and track the possible traffic incident continuously.In the aspect of object behavior recognition,this paper designs a road traffic condition and traffic incident detection algorithm for fixed camera: By analyzing the trend of the instantaneous velocity of the vehicle to determine the incidents of vehicle's ultra-high speed,ultra-low speed and illegal parking;To judge the retrograde incident by comparing the direction of the movement of the trajectory with the correct driving direction of the specified lane;To judge the illegal lane change incident by comparing the fluctuation of the horizontal distance between the track and the fixed lane line.To realize the real-time detection of the road traffic condition with the road statistic average speed as the conversion condition.Accurately identify the traffic incident types and traffic conditions.Through researches and simulation experiments of three aspects above,it is shown that the traffic incident detection algorithm based on visual saliency designed in this paper can accurately detect all types of traffic incidents and traffic conditions in real time,which is more practical and feasible.