Research On Traffic Flow Parameters Extraction Based On Satellite Video

With the appearance of high-resolution satellite video,satellite video provides a new data source for the construction of intelligent traffic systems.Obtaining large-scale road traffic information from satellite video has become a potential research direction.This paper aims to extract traffic flow parameters by analyzing satellite video image characteristics.The prime contents of this paper are as follows:(1)This paper generalizes the development of video satellite and its application,analyzing the advantages and disadvantages of satellite video as a new traffic information source,and summarizes the advantages and disadvantages of the basic algorithms involved in video traffic information collection,including vehicle detection and vehicle tracking which confirms the research direction of this article.(2)Aiming at the problems of the moving targets,such as complex backgrounds,large quantities and unobvious characteristics,a video preprocessing framework including region of interest and histogram transformation is designed to enhance the image.(3)Aiming at the characteristics of satellite video,this paper designs a method of traffic flow parameter extraction based on satellite video.The method mainly includes three parts.Firstly,taking into account that the number of vehicles pixel in the satellite video are few and the characteristics of vehicles are not obvious,this paper choose the corner detection method to achieve the vehicle detection.Different from the common corner detection algorithm,this paper adds some limitations to the traditional corner detection algorithm for the characteristics of satellite video.Through these additional conditions,it realizes the one-to-one correspondence between vehicles and corner points,so as to achieve the detection of the vehicle.Then the algorithm chooses the sparse optical flow method to track the corner detected in the vehicle detection.Due to the corresponding relationship between the corner and the vehicle,the tracking of the vehicle is realized.Finally,on the basis of the corner information obtained by the vehicle detection and tracking,combined with the relevant parameters of the satellite video,vehicle speed calculation on the road is completed.(4)This paper uses the UrtheCast HD Boston satellite video(Boston and Las Vegas area)obtained to verify the reliability of the satellite video traffic information collection method proposed in this paper.And with the help of OpenCV library to develop the program required by this paper.The experimental results show that the detection accuracv is 86.14%and the accuracy of traffic flow speed extraction is 86.80%,which proves the reliability of the proposed algorithm.