Research And Implementation Of Vehicle Collision Warning System Based On Intelligent Street Lamp Data

Traffic safety has become an important issue in the process of social development and a major factor affecting people’s sense of security^[1].The number of people who die from traffic accidents is very large in the world every year.Traffic accidents have become the main cause of abnormal death of human beings,and vehicle collision is the most common traffic accident.At the same time,the research shows that giving drivers a warning message before an accident can well avoid the occurrence of accidents.Therefore,designing a vehicle collision warning system to reduce the occurrence of vehicle collision accidents has high research value and practical significance.Data collection and transmission is the premise of system implementation,vehicle detection and tracking technology is the basis of collision warning system implementation,is also one of the key technologies in the field of automatic driving.At present,there are many problems in the vehicle detection and tracking algorithm.The extreme complexity of traffic environment leads to the change of light intensity and the occlusion of vehicles,which greatly affect the accuracy of the algorithm.This thesis explores these two problems and how to build a vehicle collision warning system.Data is collected through cameras and processed by algorithms,and the obtained data is transmitted through the intelligent light pole construction network.Data fusion technology is used to build a vehicle collision warning model and a software platform for information interaction.After testing,the system achieves the expected goal well.To sum up,this paper mainly carries out the following work:1.Build a depends on the data collection of Internet network,to integrate the camera in wisdom light pole,the wisdom light pole deployed in a high incidence of accident stretch of road intersection with surveillance cameras and vehicles on the road of on-board camera build a data collection network,through the perspectives of data collection can better response of the vehicle state.2.The data transmission network is constructed.By comparing several wireless network technologies,the transmission information is divided into equipment state information and vehicle characteristic information,and the sub-G and NBIOT hybrid transmission network is constructed to classify data transmission and improve data processing efficiency.3.The existing detection and tracking algorithm is optimized,and a fusion target detection method is designed.After experimental verification,the accuracy of the method has been improved to some extent.An improved tracking algorithm based on HLBP feature fusion Kalman and Camshift is proposed,which has good effect under the condition of light intensity change and partial occlusion.4.The feature information obtained through the algorithm and network is screened,analyzed and fused,and the vehicle feature information is combined with the vehicle track tracking information to build a vehicle collision warning model for danger warning.At the same time,the suspected traffic accident occurred in remote sections can be processed,and the information is sent to the hospital as soon as possible after confirmation.The realization of drivers have a clear understanding of their own driving state,as far as possible to avoid the occurrence of traffic accidents.5.Design a vehicle collision early warning software platform with software development technology,realizing real-time interaction between drivers and information,and better obtaining early warning information and managing various equipment through the software platform.