Research On The Recognition And Early Warning Of The Target In Front Of The Intelligent Vehicle Based On Deep Learning

Nowadays,China’s automobile production and sales have ranked first in the world.While cars bring convenience to people’s life,they also cause problems such as traffic accidents,air pollution and traffic accidents.The number of casualties caused by traffic accidents is increasing year by year.With the continuous progress of artificial intelligence and other related technologies,as well as the application of cameras,millimeter wave radar,lidar,automatic driving controller and other devices in automobiles,people hope to assist or replace drivers to drive vehicles through technical means to improve the safety of vehicles.The key factor of vehicle driving safety is the target recognition and early warning,which is usually carried out by camera,millimeter wave radar and other equipment.This paper is positioned in the field of intelligent vehicle forward target recognition and risk warning.Based on the monocular vehicle camera,the paper conducts theoretical and experimental research on target location,ranging and risk warning by combining the neural network technology,aiming at the current problems such as slow target recognition speed,low target positioning accuracy and large target ranging error.This paper firstly expounds the research status,development trend and policy of intelligent vehicle at home and abroad.Then,the achievements of domestic and foreign researchers in object detection based on machine vision are listed,and the significance and necessity of this paper are discussed.After that,the image preprocessing technology is compared and discussed extensively,FAST algorithm is selected for target detection,and the algorithm is optimized.Then,the principle of camera calibration and ranging is studied and analyzed,and a new target ranging algorithm based on coordinate transformation is proposed.Then,the principle of neural network are introduced,and the different methods of target recognition and feature point detection and analysis,demonstrates the intended to two different vehicles and pedestrians,two different ranging method is proposed,the vehicle target presents a new optimized FAST feature point detection algorithm,and achieved good effect of target recognition,which laid a foundation for improving the precision of target range.Then,based on the automobile braking theory and by comparing the braking distance with the target to calculate the distance,a risk early warning system was established on MATLAB 2016 B platform to warn the risk targets ahead.Finally,the effect and accuracy of target detection and ranging are verified through experiments,and the effect of the driving risk early warning system designed in this paper is verified.The test results show that the proposed target detection and ranging method is feasible,and the designed risk early warning system is accurate and effective.In this paper,the machine vision technology has been studied and tried in a beneficial way.The research results have certain practical significance for the fast identification and early warning of the risk targets ahead,and provide support for the application and promotion of the machine vision technology in the future intelligent vehicle environment perception technology.