Research And Design Of Driver's Dangerous Driving Detection System Based On Machine Vision

In recent years,the sharp increase in the number of cars has made traffic accidents more and more frequent,which poses a great threat to people's lives and socio-economic development.The related research shows that driver factors dominate the cause of traffic accidents,So it's of great scientific significance to study the relevant technologies of driver's dangerous driving behavior detection.The main goal of this project is to design a detection system that monitors the driver's dangerous driving behavior and gives real-time warning.Based on the analysis of the development status of dangerous driving detection technology at home and abroad,the paper studies and designs a dangerous driving behavior detection system based on machine vision.The paper introduces the relevant theories and key technologies involved in the system.This article analyzes the basic concepts and advantages and disadvantages of several commonly used face detection algorithms.Focused on the face detection algorithm based on Ada Boost.The working principle of the face detection algorithm is described from the aspects of Haar feature extraction and Ada Boost classifier design.In order to realize the function of distracted driving recognition,paper studies the principle of image recognition using convolutional neural network,providing a theoretical basis for subsequent software design.Aiming at the feature of the human eye that best reflects the degree of fatigue,the related algorithms of human eye positioning and tracking have been studied.Finally,an algorithm combining Hough transform and gray-scale projection was selected to achieve accurate positioning of the driver's eyes.According to the function and performance requirements of the system,a hardware development platform with DSP chip TMS320C6748 as the core processor is designed.The system is divided into five modules: image acquisition,CPU processing,external storage,power supply,and voice alarm for hardware design.The hardware design is mainly realized from the two aspects of key chip selection and circuit design.According to the software design goals of the system,software design is carried out from the aspects of fatigue driving detection and distracted driving recognition.For the software implementation process of fatigue driving detection,research and design a detection method based on the four steps of image preprocessing,face detection,human eye positioning and tracking,and fatigue driving judgment.Using MATLAB and Open CV function library to simulate related algorithms,verifying the feasibility of the software implementation process.Aiming at the software implementation problem of distracted driving recognition,according to the operating principle of Alex Net neural network,a method was designed to use a large amount of image data to train the network to identify distracted driving behavior in sample images.Finally,this article builds a DSP-based dangerous driving detection system based on a combination of hardware and software design,And implement software writing in DSP / BIOS operating system.An experiment was designed according to the performance requirements of the system.The experimental results show that the accuracy of the system in detecting dangerous driving behavior of the driver under normal circumstances reaches about 82%,and the system has good real-time performance.