Research On Fatigue Driving Detection Of Bus Drivers Based On Image Processing

With the improvement of people’s income level,the number of vehicles is increasing day by day,road traffic accidents have become the most important factor of global casualties,the number of domestic road traffic deaths is also increasing year by year,the traffic safety situation is becoming more and more serious.It is found that the driver’s fatigue driving behavior is one of the main factors affecting the road traffic accident mortality,and the driver’s behavior is controllable.In order to reduce the traffic accidents caused by fatigue driving,it is very important to study the fatigue driving warning system.In this thesis,the technology of fatigue driving detection based on image processing is studied.Through the in-depth study of face location,face feature point location,eye feature point location,eye fatigue feature extraction and fatigue judgment,a real-time fatigue detection system for drivers is implemented based on this theory,which can detect driver’s fatigue better and send out fatigue in time early warning.The main work of this thesis is as follows:(1)Face detection.Integrating open-source face da Table ase samples,combining Ada Boost with Open CV,in order to reduce the missing detection caused by face tilt,a new Haar-like feature is added to train the face detector,and the trained face detector is tested.(2)Face feature point location and eye location.On the basis of face location,the face feature point locator is trained by using random forest regression model.The test results show that the feature point location of face is accurate,and the feature point location of human eye can be obtained,which provides support for the feature extraction of human eye fatigue.(3)Eye fatigue feature points extraction and fatigue judgment.After the facial feature points are located,the feature points of human eyes can be located.The fatigue features of eyelid opening and closing degree are extracted and the corresponding fatigue eigenvalues are calculated.Compared with the traditional PERCLOS fatigue judgment method,this thesis proposes a new fatigue judgment method based on support vector machine,and compares the two methods by experiments.(4)Fatigue driving detection program based on C ++.C ++ builds the fatigue detection and data writing system,captures the network camera,detects the fatigue and records the fatigue information.The experimental results show that the system has a good accuracy when detecting fatigue,and it will send out fatigue warning in time.Finally,a Java based management platform is built to read driver fatigue driving data for future statistical analysis.