Research On Driver Fatigue Detection System On The Embedded System

In order to reduce traffic accidents rate and ensure the safety of people’s life and property,the effective detection and prevention of driver fatigue driving has important researchsignificance.The paper implements a fatigue driving detection method on embedded system whichbased on machine vision and PERCLOS fatigue decision algorithm and transplanted into theembedded Linux system. The method has achieved vehicle, non-contact, low power andensured real-time and high portability. The major work of this thesis is as follows:1. The paper has built the hardware platform and software development environmentfor driver fatigue detection on embedded system. The embedded system wasimplemented image data real-time acquisition by USB camera and subsequent imageprocessing functions.2. In order to meet the needs of Image processing and face detection on embeddedsystem platforms, the paper has compared the existing face detection methods.Thepaper has achieved the driver’s face detection and position on the Complexbackground image, the method is based on the Adaboost algorithm by Haar-Likefeatures.3. The paper is based on the Adaboost algorithm for eye detection and image areabinarization to implement the driver’s eye detection and position. The fatiguedetection system uses PERCLOS fatigue determination to count the closed eye tocomplete fatigue detection on embedded system.The fatigue detection system has completed the practical requirements for fatiguedetection on embedded systems. The practical result is consistent with the PERCLOSdetermination. The method is simple and computing speed quickly, the recognition andrecognition rate is higher.