Research On Moving Vehicle Detection Based On UAV Videos

With the development of unmanned aerial vehicle(UAV)technology,the development and application of UAV in the field of civil industry has become a hot research topic.UAV has the advantages of low cost,easy deployment,high mobility,large viewing range,uniform scale and etc.In this paper,we consider the different hover and cruise flight status of UAV in the process of video capture.According to the characteristics of UAV video images,a moving vehicle detection algorithm based on UAV video is designed and developed.The experimental results are compared with the frame difference method,the hybrid Gauss algorithm and the optical flow method.Experimental results show that the proposed algorithm has high detection accuracy and can meet the needs of real-time applications.The main contents of this paper are as follows:(1)This paper analyzes the UAV's development prospects and industry application development advantages,summarizes application research status in the field of transportation machine,sums up the relevant motion platform video object detection technology and image processing technology,and analyzes the mainstream algorithm of moving target detection based on video in detail.(2)In consideration of the different hover and cruise flight status of UAV,UAV videos have different processing needs.We develop a moving vehicle detection algorithm based on UAV video.The algorithm is divided into three parts.First of all,we set the region of interest to select the range of moving vehicle detection.Then the ViBe background updating algorithm is introduced to extract the moving foreground region,which improves the defects of the ViBe algorithm using the fixed threshold and the ghost problem.Finally the detection of moving vehicles is accomplished by morphological optimization and vehicle target recognition.(3)Based on the qualitative and quantitative analysis,three kinds of moving object detection algorithms,the frame difference method,the hybrid Gauss algorithm and the optical flow method,are compared with the algorithm in this paper.Experimental results show that the proposed algorithm has high detection accuracy(90.25%)and real-time computation speed(42.69fps).The algorithm can effectively detect the moving vehicle based on the UAV video and realize the real-time application.This research has important reference significance for enriching the vehicle detection algorithm and the traffic information collection method based on UAV video.which is helpful to the application of UAV in traffic field.