Research On Road Environment Recognition Of Intelligent Vehicle Based On Binocular Vision

With the rapid development of the automobile industry and the improvement of the social and economic level,the automobile has become more and more popular all over the world.With the rapid development of the automobile industry,the radius of human travel has been improved,which makes the transportation more convenient,but the rapid development of the automobile industry has brought a lot of hidden dangers to human beings and society.For example,vehicle exhaust pollution to the environment,too many vehicles will make the city traffic congestion and traffic accidents caused by vehicles and other issues.The effect of frequent traffic accidents brought to the security of the human society is the most serious,how to effectively improve the focus of traffic safety has become international researchers.So the concept of intelligent vehicle came into being.In the research of intelligent vehicle,the development trend of intelligent vehicle is based on the combination of multi sensors(laser radar,camera,infrared detectors)and vehicle.The research and development of machine vision has become the focus of today’s research,while binocular vision is an important branch in the field of machine vision,it is through the simulation of biological machine vision to achieve the perception of the environment.In this paper,the method of binocular stereo vision is used to realize the extraction of the road lane and identification of obstacles,the measurement of the target distance and the construction of the minimum safe distance model.The main research contents are divided into four aspects:(1)Build experimental platform.In this paper,a movable horizontal slide rail is designed and installed on the laboratory electric vehicle,and the binocular camera is installed on the slide rail.Through this platform to complete the image acquisition.(2)Research the lane and obstacle detection algorithm.The experimental images collected on the platform are preprocessed.And then through the improved HOUGH transformation of the lane line to identify the extraction.According to the detection results to extract the ROI area,the region for edge detection and image entropy calculation to find the approximate area of the obstacle.And then the feature points are extracted by SIFT operator to find the exact position of the obstacle.In this paper,we propose to use the Kalman filter to track the target,and combine the NMI feature and the entropy to verify the prediction area to achieve the goal of improving the tracking accuracy.(3)Research on Ranging Algorithm Based on Binocular Vision.Using binocular vision system for distance measurement experiments.According to the principle of binocular imaging of the collected images and extract the parallax value to complete the calculation of obstacle distance.And in the process of distance measurement through a large number of experiments found at different distances under the baseline length on the measurement accuracy of the impact.To achieve the purpose of reducing the experimental error rate.(4)The minimum safety distance model of the car is established.First of all,analysis the safety warning system and the car brake process,and then according to the principle of braking to establish the minimum safety distance model of the car,Finally,this model is simulated in MATLAB,and the experiments prove the model achieves the expected effect.