Driver fatigue detection system based on DSP

Abstract

To detect driver fatigue states effectively and in real time, a driver fatigue detection system was built, which take ICETEK-DM6347 module as system core, near-infrared LED as light source, and CCD camera as picture gathering device. An improved PER-NORFACE detection method combined several simple and efficient image processing algorithms was proposed, which based on principle of PERCLOS method and take the human face location as the main detection target. To ensure the ability of real-time processing, the algorithms on the DM6437 DaVinci processor were optimized. Experiments show that the system could complete the driver fatigue states detection accurately and in real time.

Real time detection of driver fatigue based on CNN‐LSTM

“…This method uses visual sensors to obtain the driver's head movement [29,31,32], eye state [18][19][20][21][22][23][24], mouth state [25,26], and facial features [27,28] etc., and extract the corresponding driver behaviour features from the above data, and realize the classification of fatigue status according to the features and the detection model.…”

“…Wang et al [19] have proposed an improved PER NORFACE face detection method, based on PERCLOS method to detect the fatigue state of the driver. Orazio et al [20] detected the opening and closing of the driver's eyes in the image sequence, and used the probabilistic model to evaluate the time incidence of opening the eyes, and then identified abnormal behaviours such as driver inattention or drowsiness.…”