Research On Detection Algorithm Of Fatigue Driving Behavior Based On Video

With the rapid development of China's transportation industry,cars have long become the necessary means of travel for people's traveling ? While enjoying the convenience of traveling by car,people are bothered by the security risks brought by the car.In all traffic accidents,most accidents are caused by drivers' fatigue driving.In the advanced information technology,it is of great practical significance to accurately judge the driver's fatigue state and give the alarm in the course of driving.This method based on video processing,first to detect the face information in the video driver,detected in the driver's face based on the information,feature points matching up to the driver's eyes and mouth,through the analysis of the driver's eyes and mouth features,to judge the driver's mental state from a plurality of parameters,thereby accurately and timely to detect the fatigue state of the driver.The specific work and innovation of this article are as follows:(1)In order to realize the monitoring of driver's driving in the daytime and nighttime,we use near-infrared camera to collect the driver's driving state,so that drivers can monitor during daytime and night driving.(2)Firstly,the video image is preprocessed.Then the optimized Adaboost algorithm is used to locate the input video.After locating the driver's face information,we use the optimized SDM algorithm to further extract the feature points of the driver's facial features.In the driver's facial feature points,the state changes of the mouth and eyes can best reflect the driver's mental state,and the driver's eyes and mouth information can be extracted through the matching of SDM feature points.The test results show that the results of the optimized Adaboost algorithm and the SDM algorithm have good real-time performance and robustness.(3)For more accurate and comprehensive driver fatigue detection,we use PERCLOS,blink frequency and yawn frequency to detect fatigue for drivers.In view of the limitations of the PERCLOS algorithm,the blink frequency parameters are introduced,and two parameters are used to describe the information of the driver's mouth.The yawn frequency is used to describe the driver's mouth feature.The three parameters work together to improve the accuracy of fatigue driving test.(4)Using C++ programming language,based on OpenCV and QT development environment,we design a PC based fatigue driving behavior detection software,and implement the algorithm in this paper.