Research Of Fatigue Driving Detection Algorithm Based On Machine Vision

Fatigue driving is one of the most important factors in traffic accident and the first causein traffic accidents death, with the increasing number of the vehicle, fatigue driving hasbecame a serious social problem. Therefore, the research of fatigue driving detection systemhas a great significance for enhancing traffic safety and safeguarding the security of people’slives and property.In this thesis, lots of fatigue driving detection methods were researched and analyzed andone method which is based on machine vision was introduced. The major work of this thesisis as follows:1. Making an overview of various face detection methods and realizing a face detectionmethod based on skin color. This method uses YCbCr color space to establishtwo-dimensional gauss mathematical model based on the color distribution and locates thedriver’s face by using skin color similarity and face shape characteristics together.2. Realizing the face detection method which is based on AdaBoost algorithm, analyzing andcomparing skin color model algorithm and AdaBoost algorithm which was selected as thefinal solution of face detection because of having better effect.3. This thesis introduced an eyes location method using a combination of binarization andgray integral projection. Besides, it compared this method with another eyes locationmethod which is based on eyes’ edge feature in order to prove the former is more feasible.Finally, it used PERCLOS algorithm to calculate driver’s eye closure frequency anddetect fatigue driving.