Research On Driver Fatigue State Detection Method Based On Video

Fatigue driving is one of the major hidden dangers of many traffic accidents.With the rapid development and application of machine vision technology,the use of non-contact,machine vision based fatigue detection technology has become the mainstream research direction in this field.At present,machine vision based fatigue detection method is difficult to do well in both accuracy and speed.To solve the problems above,the fatigue state detection based on video is developed by the human eye state,and the driver's fatigue state is detected and judged.The main contents and related conclusions of this paper are as follows:Aiming at the influence of illumination factors on face image segmentation and location,clustering characteristics on chromaticity space by using skin color,a face location method based on skin color segmentation is adopted.The Gauss skin color model is established on the YCbCr color space for color face images,and the color similarity of the skin color is calculated to enhance the difference of the color area and the background.After a series of image processing works,we can locate the face region by gray level integral projection.This method is less affected by light factors and has high recognition rate.An iterative cascade classifier combined Gentle Adaboost and Haar-like rectangular features is adopted to solve the problem of instability in closed eye position.The detection of human eye fatigue based on video needs to capture continuous frame dynamic face images.For the human eye tracking and positioning of continuous frames,a comprehensive detection method based on Calman filter and improved Camshift is used to complete the dynamic and real-time tracking detection of human eyes.The results show that this method can effectively locate and track human eyes in dynamic face images.Aiming at the problem of low accuracy of human eye recognition,a new algorithm based on the minimum circumscribed rectangle is proposed to get the information of human aspect ratio.By transforming the width to height ratio of each eye sub image to the opening and closing information of human eye,the corresponding mathematical transformation relation is adopted.By calculating the PERCLOS value and blink frequency,we can judge driver's fatigue state.Finally,a video based driver fatigue state detection system is implemented,and a lot of simulation experiments are carried out under PC.The related experimental results show that the system has fast computation speed and high detection accuracy,and has certain practicality.