Research On Vehicle-assisted Safe Driving System Based On Infrared Video Image

With the gradual improvement of people's living standards,the number of vehicles on the road is increasing,and traffic accidents are also on the rise,posing a serious threat to people's lives and property.Especially in the low-light environment,the driver can't correctly judge the obstacles in the front sight distance,which leads to a significant increase in the risk of driving.Therefore,research on vehicle-assisted driving technology,how to apply computer information technology to assisted driving to improve the safety of driving,has been the focus of attention of major auto manufacturers and research institutions.Because the far red hot camera is not affected by the light,it can rely on the thermal radiation energy emitted by the object itself to perform imaging in the case of poor visibility environments such as rain,snow,fog,and darkness,compared to visible light imaging infrared thermal imaging.Technology in the dimly-sighted scene can greatly improve the driving safety.This article uses the Cortex A8 series MY-IMX6 development board introduced by ARM as the system controller and the far infrared thermal imager FLIR-A310 as the infrared video acquisition equipment for the driving environment under adverse conditions of rain,snow,fog and other fields of view.LCD liquid crystal display as a display device to build a hardware development platform.With embedded Linux realtime operating system,Caffe neural network framework,OpenCV3 machine vision library and Qt5 graphical interface library as the system software development platform,a set of driving assistance safety driving system was designed and implemented.The system incorporates infrared thermal imaging,image processing,deep learning and computer vision technologies to detect common obstacles such as pedestrians,vehicles(cars),and animals that appear in the sight of rain,snow,fog,and night,etc.Dogs)are detected and given a dangerous warning to the driver within the effective range of driving.This article takes the infrared image of the vehicle under the embedded system as the main research object,to carry out the infrared image enhancement processing and the night vision and pedestrian,vehicle,animal and other multi-target detection research.In driving,the system can greatly improve the safety of driving at night and in special weather conditions,and has a high practical value and research significance for vehicle-assisted driving.