Forward Vehicle Detection And Ranging Algorithms Based On Deep Learning

With the rapid development of China's economy and industry,people's living standards have been gradually improved.More and more families have purchased cars,and the number of cars in our country has increased year by year.As a result,the traffic congestion in urban transportation has been aggravated,resulting in more and more bad traffic environment and traffic rear-end collision accidents.This problem can be solved by Intelligent Transportation System(ITS).The more important component of the system is Intelligent Vehicle.It is necessary to detect and identify the vehicle in front,calculate the distance between the vehicle and the vehicle in front,and maintain the appropriate safe driving distance with the vehicle in front,so as to give the driver a certain amount of time to take appropriate measures to reduce the incidence of rear-end collision with the vehicle in front.Therefore,the detection and ranging of the front vehicle is the core part of the intelligent vehicle,which ensures the safety of the vehicle in the process of driving.In this paper,the application of deep learning in image detection is studied in combination with the deep learning proposed in recent years.At the same time,Based on the research of traditional geometric distance measurement model based on monocular vision and the idea of mathematical regression modeling,a distance measurement model based on radial basis function neural network is proposed.This paper mainly studies the forward vehicle detection and ranging based on deep learning.The main contents of this paper are as follows:Firstly,the research background and status quo of vehicle distance measurement technology in the front are described.The existing vehicle detection algorithms and distance measurement algorithms are summarized and summarized.The advantages of the vehicle detection algorithm based on machine learning and visual distance measurement algorithm are illustrated.Secondly,the algorithm of vehicle detection based on machine learning is studied.The algorithm steps of vehicle detection based on Adaboost and the algorithm of vehicle detection based on depth learning are described in detail.The experimental results of the two algorithms are compared.The advantages of deep learning over Adaboost are analyzed from the accuracy and running time of vehicle detection.Then,the forward vehicle ranging algorithm based on computer vision is studied,and the traditional geometric imaging model based on monocular vision is described in detail.Considering the influence of imaging system error,lens distortion of camera,error in camera calibration process,and road geometric constraints,the idea of distance measurement based on data regression modeling is used to make use of diameter.To solve the above problems,the hidden layer of radial basis function neural network(RBF-NN)is proposed,and a ranging model based on RBF-NN is proposed.Finally,the forward vehicle ranging algorithm based on deep learning is studied.The candidate regions are obtained by using the multi-scale anchor points and the regression output layer of Faster R-CNN through the region recommendation network.Then the fast region convolution neural network processes the candidate regions to detect the front vehicle in the image.Finally,the edge information of the front vehicle is input into the radial basis function.The distance measurement model trained by neural network is used to get the distance of the vehicle in front.The KITTI database images are used to test,which can be verified by experiments.The forward vehicle ranging algorithm based on depth learning proposed in this paper can meet the basic requirements of the forward vehicle ranging.