Research And Design Of Vehicle Security Distance Sensing System Based On CNN

CNN is a convolutional neural network and is a deep feedforward artificial neural network.It can extract high-level features of images and has been successfully applied to the classification and recognition of images.It has great influence in the field of computer vision.In recent years,rapid economic development has gradually improved the people's material living standards,and has also increased the level of construction of urban and rural highways.Car ownership has shown a trend of significant growth compared to the past.The speed of traffic on the highway is also increasing,and the pressure on urban traffic is also increasing.The drastic increase in traffic pressure has caused a large number of traffic accidents,which has brought a lot of threats to people's travel safety.It is imminent to reduce the incidence of traffic accidents.The important means to solve modern traffic problems is the intelligent transportation system,and the active safety technology of the car is an important part of the intelligent transportation system.Its advanced driver assistance system has been increasingly valued by the society in the course of continuous improvement and has become one of the main driving force for the development of automotive electronics in the future.In the process of driving the vehicle,the safety distance is a necessary condition for personal safety.If the car can automatically identify the vehicle ahead and determine whether the distance is safe,the number of traffic accidents caused by the rear-end collision will be greatly reduced.Therefore,research on vehicle safety distance detection technology has important practical significance for improving road traffic conditions and improving driving safety.This thesis compares and analyzes various current vehicle safety distance detection technologies.Based on machine learning and convolutional neural networks,it studies the vehicle safety distance sensing system and proposes a vehicle safety distance detection method based on deep convolutional neural network.The vehicle safety distance sensing system has intelligent sensing capabilities,which can correctly identify the type of vehicle ahead and determine whether the distance traveled is safe.This thesis focuses on the principle and method of object detection and digital image processing,preprocessing the acquired image data,such as gray-scale transformation,histogram equalization and image denoising,so that the convolutional neural network can be better extracted and analysis data characteristics.The thesis focuses on the working mechanism and design method of machine learning,convolutional neural networks,and design a deep convolutional neural network to realize the perception of vehicle safety distance.Using the collected driving environment data to train the convolutional neural network and test its distance perception effect.According to the verification results,the network control parameters and the network structure are continuously adjusted to make it have better detection accuracy and work efficiency.This paper mainly tests the target detection,image pre-processing and image recognition modules in the vehicle safety distance sensing system.Through programming,it realizes the functions of each module in the system and builds a software testing platform for simulation testing.Experimental results show that the target detection module can detect the front target vehicle and extract the image ROI in real time under complex background environment.The image preprocessing module not only can effectively reduce the amount of data,enhance the detectability of important information in the image,but also can improve the reliability of image feature extraction,classification and recognition,and has a faster processing efficiency.The image recognition module can realize the recognition of the front vehicle type and the perception of the vehicle safety distance,and has relatively good detection accuracy and response sensitivity.The accuracy of the vehicle safety distance detection can reach 97.5%.It has positive significance for improving the driving safety of vehicles,improving the existing road traffic conditions and reducing the occurrence of traffic accidents.