Driver's Face Image Recognition for Somber Surroundings Based on Computer Vision

摘要

A CCD camera was installed in the cab to monitor the driver if he/she fatigue or drowse, especially at night. The technology of driver fatigue monitoring plays an important role in reducing the rate of traffic accidents. It is difficult to recognize the driver's state because the facial image is fainter under somber circumstance. A new method is proposed to detect and location driver's face in order to judge if he/she is drowse. By using the method of combining image corrosion and image expansion get rid of the image noise. Then strengthening facial grey value through nonlinear curve normalized to obtain a proper gray value. Project the facial image to horizontal and vertical to get the driver's face location. It is the key to monitor the driver. The method can be applied in dusky light to locate the face, effectively remove the shadow around interference. Using Visual C + + 6.0 development a program and a test was carry under different environmental and road. The results show that, under the somber light, the driver faces positioning accuracy rate is 87.3% with the property of real-time and high accuracy. This method provides a sound foundation to further identify the driver's fatigue at night.