Real-time Dsp-based Driver Fatigue Detection System

According to the transport sector statistics, our country is the first in the world that the highest number of deaths in road traffic accidents has been for several years. Drivers' fatigue driving is a major cause of traffic accidents in many countries, which are actively researching driving fatigues detection. Therefore, how to effectively monitor and prevent driver fatigue driving has much real significance to reduce traffic accidents and personnel mortality.A set of real-time fatigue driving detection system based on DSP is designed under the relevance of the various fatigue and fatigue characteristics; the PERCLOS (eyes closed time unit time percentage) is selected as the basis to determine the extent of fatigue, Driving fatigue measuring system based on DSP that vehicle mount, non-contact style, real-time system is brought forward using machine vision and image processing methods to detecting changes in the eyes of information.Firstly, muscle fatigue and mental fatigue are analyzed from the physiology perspective, and studied the physiology of driving fatigue mechanisms. Especially studies focus on the eyes fatigue, from the US Federal Highway Administration FHWA (Federal Highway Administration), the study of research literature driving fatigue, discussed PERCLOS and other eye measurement of the effectiveness of activities, Recognizing fatigue measurement vehicle driving is the best way to express their fatigue level using PERCLOS as indicators.Then video collection system is composed of CCD, TMS320DM642, TP5150, SAA7121 chips for fatigue detection algorithm because of data volume, high-speed transmission, the real-time needs of complex operations. The data collection system used camera control to meet the real-time multi-way video acquisition, processing and displaying needs, it could serve as fatigue detection algorithms, video processing and image processing hardware platform.Finally the face and eyes of the real-time positioning and detection algorithm are designed. Which use skin colour model based on the face characteristics and difference frame combining algorithms for human face real-time position and detection.the paper design a new approach - cumulative histograms threshold method for human face binarization, we can keep our eyes from the face colour separated and reduced to illumination sensitivity, and, then use the connectivity domain search algorithm from nothing extent around the eyes to avoid fragmentation and improve the calculation speed. Then the eyes are accurately positioned using eye image for extracting eyes from human face rapidly and exactly. Finally the fatigue degree is judged respectively Perclos and short durationod eyes two parameters.Many tests show that the system can detect blinking times and visible pupil areain real time. So it can det(?)t the degree of fatigue efficaciously.