Driving Safety Detection Technology And Application Integrating Vehicle Trajectory And Machine Vision

The vigorous development of mobile communication technology,global positioning technology and Internet of Things technology has given birth to massive amounts of traffic data.By collecting the state data of "people-vehicle-environment" in the process of road transportation,and using machine intelligence to conduct in-depth mining and fusion analysis of these data,it is helpful to discover the hidden traffic behavior characteristics of the vehicle,especially the detection of the possible traffic of the driver Illegal behavior is of great significance for preventing and reducing the occurrence of traffic accidents.To this end,this article comprehensively considers the traffic data of the driver,the vehicle and the environment,uses the bionic neural network technology to detect anomalies in the trajectory data,uses the machine vision technology to analyze the status of the environment and the driver data,and integrates the three analysis results Form a more comprehensive vehicle safety detection technology to provide a reliable basis for the detection of vehicle traffic violations.The main research work of this paper includes the following three aspects:(1)In order to realize the detection of the running state of the vehicle trajectory,a vehicle trajectory prediction method based on the bionic neural network is proposed for the time series characteristics and position relationship of the vehicle trajectory.The trajectory is predicted by the long-short memory neural network(LSTM),and the trajectory prediction results are analyzed according to the trajectory characteristics to achieve the purpose of identifying abnormal trajectories.In the trajectory prediction,the improved bionic firefly algorithm(FA)is used to optimize the parameters of the prediction model,which improves the processing efficiency of the model,enhances the data fitting ability of the model,and thereby improves the accuracy of the prediction;(2)Use machine vision technology to establish a road environment and driver behavior analysis model to effectively identify the road environment and driver behavior.On the one hand,the effective ROI(Region of Interest)area of the environmental identification is extracted through the image preprocessing technology,and the multi-class support vector machine is used to complete the identification of the environmental identification.On the other hand,a human-hand-posture estimation model is established based on Open CV to extract eye,mouth,hand,body movements and other behaviors in the driver sample,and identify dangerous driving signals such as squinting,yawning,smoking,and calling,To detect unsafe driving behavior;(3)Research the method of establishing a driving safety evaluation model,apply information fusion technology to the evaluation of driving safety level,and comprehensively evaluate the driving safety status from the three dimensions of vehicle trajectory,traffic signs,and driver characteristics.And through the combination weighting method,the weight distribution of the three indicators is carried out,and the driving safety assessment result that is closest to the actual state is obtained.