Research And Implementation Of Traffic Detection Method Based On Vehicle Identification

Intelligent video process has been widely applied in “intelligent traffic”,“wisdom home” and so on.Through the analysis of video surveillance,we can effectively manage traffic and ensure the safety of the people's property.Traffic detection is one of the focuses in the field of intelligent video analysis.However,traffic detection is currently facing the problem of slow detection and low accuracy.Traffic detection based on vehicle identification is designed to improve the detection speed and accuracy through vehicle identification module.In order to improve the processing rate of traffic detection,reduce the detection rate of false positives,get better results in the application of the real scene,the paper launched an in-depth study in:(1)Research on the method of moving target detection combined with background updateing and shadow removing.In this paper,we use Gaussian model to build the backgournd model and update the background.By using background difference method and three frames difference method to extract foreground.At the same time,to reduce the vehicle error detection,vehicle adhesion,vehicle changes,etc.,caused by the shadow,we realize the shadow removal algorithm based on HSV chromaticity space.Improve the processing speed by reducing the detected area.(2)Research how to realize vehicle feature extraction with the multi-feature fusion method.This paper fusion LBP features,color moment features,Hu moment features,circularity and angle features,and Fourier descriptive features,then train the classifier.Through the multi-feature,we can improve the capability of the classifier.(3)Reserch on the target tracking algorithm.Through this algorithm we can record the position of the vehicle in the subsequent video sequence.(4)Reserch on the method of traffic statistics combined with target tracking and virtual detect line.First,we mark the vehicle position by the target tracking algorithm,then determine whether the vehicle through the virtual test line,only when the vehicle through the test line we count and get traffic flow.This paper focuses on the technology of moving target detection,feature extraction and classifier training,and then designs the traffic detection method.On this basis,complete the realization of traffic detection system based on vehicle identification.Finally,test the software system functions and performance,Form the test results,the traffic detection method based on vehicle identification constructs a classifier with excellent performance,improves the processing speed and the accuracy of traffic detection.