Research On Forensic Technology Of Illegal Vehicle Behaviors Based On Imaging Control

With the development of economy and society,the transportation industry in China has made great progress.The developed transportation network has provided great convenience for people’s daily life,and has also had great potential problems.At the same time,the security industry is rising rapidly.Surveillance technology based on IP cameras and other monitoring devices proposes a very effective way for traffic safety supervision.This paper does some research on the technology of forensics for the vehicles’ lawbreaking deeds focusing on the traffic monitoring field.And then based on ONVIF protocol,adaptive vehicle recognition algorithm and improved vehicle tracking algorithm,the relevant functions of the entire system are realized,such as video collection,vehicle recognition and tracking,vehicle violation detection and snapshot of the illegal vehicles.The main contents of this paper are as follows:Research of image acquisition technology based on ONVIF protocol.Based on ONVIF protocol,the basic universal interface of network video monitoring software is realized in this paper.First,it analyzes the work mode and main functions of ONVIF protocol,such as equipment search,equipment management,equipment IO and other functions.Then the implementation mechanism of SOAP is analyzed and the foundation for the implementation of ONVIF interface has been laid with the work such as parsing the wsdl files and compiling framework code.Then the parameters and their respective meanings required by the ONVIF protocol are analyzed.On this basis,the equipment address acquisition,media configuration file acquisition,and Uri acquisition are completed.Research on vehicle detection and tracking technology.For vehicle detection,The update efficiency of the ViBe algorithm takes a fixed value which will cause problems such as eliminating the ghost regions too slow or melting the slowly moving target into the background.This paper proposes a background change degree function with the elimination of ghost areas as the indicator by analyzing the effect of update efficiency on vehicle identification in real time process.Then based on this function,the adaptive updating efficiency ViBe algorithm is proposed.The experiment shows that the algorithm can quickly update the background sample when thebackground changing degree is abrupt,and reduce the updating efficiency to a suit scale when the background is relatively stable to prevent the foreground object to be integrated into the background.For vehicle tracking,the classical Calman filter algorithm does not consider the problem of vehicle occlusion.If vehicles occlude and two targets overlap,it will have a serious impact on the prediction of the vehicle information,such as losing the target.An improved Calman filtering algorithm for vehicle occlusionand a vehicle occlusion decision algorithm based on edge point statistics are proposed.Experiments show that the tracking algorithm can process vehicle occlusion on the basis of real-time performance.Research on detection and forensics of illegal vehicle behaviors.This paper mainly studies the detection functions for three kinds of illegal behaviors,including illegal parking,vehicle retrograde and illegal vehicle lane changing.For forensic technology,the current traffic illegal forensics usually adopt a fixed angle camera,which can only capture the illegal acts in a fixed position,and can not achieve goodresults for the illegal acts in other positions in the scene.In order to solve this problem and complete the process of obtaining evidence,this paper control s a single spherical camera to rotate and zoom in according to the detection results of illegal vehicle behaviors to collect the pictures which meet the requirements of “illegal behavior image forensics specification for road traffic safety”.The experiment shows that the illegal vehicle behaviors can be efficiently snapshotted via using the detection and control method in this paper.System software implementation.The whole system is divided into two aspects,the client and the server.The client mainly has the functions of detection mode selection,real-time video display and capture result query.The server mainly implements real-time video acquisition,vehicle detection,vehicle tracking,illegal behavior determination and cloud platform control.The communication between client and server is carried out through communication module.Finally,C++ is used to implement the server-side code,and C# to build the human-computer interaction interface.The experiment shows that the software can realize the display of real-time video,capture the illegal behavior according to the selected detection mode,and query the results of the capture.