Research On Multi-mode Driver Fatigue Detection System

With technology progressing, the automotive industry has been developing rapidly,。The population of vehicle in the world has been greatly enhanced. Traffic accidents followed by were increasing year by year. Proportion of traffic accidents due to fatigue driving causing is also growing. Therefore, this issue has far-reaching and important research significance.This paper proposes a detection method based on visual fatigue, using the camera capture driver face images. Detect the location of human eye and mouth and judge their state, so as to determine driver fatigue state. The main content and the results obtained are as follows:First, human face images were pretreated. After using the nonlinear color transform to human face, obtained skin color regions. Through judging the size of the region, exclude non-human face skin color regions to get the approximate area of face.Secondly, on the basis of locating the human face, precisely located human eye and the mouth. Using geometric features of human face, that is, eyes were in the upper half face. Combining gray integral projection and one-dimensional wavelet transform, obtained the exact location of the human eye in two steps. After locating of the eye, using the relative position of human mouth and eyes, combined gray integral projection and one-dimensional wavelet transform again and find the exact location of the mouth.Then, next is to judge the state of eyes and mouth. Using light regulation and segmentation method, the human eye and mouth features were extracted. Calculated the number of black pixel and height to judge the state of eye and mouth.Finally, using the state of eyes and mouth in the image to judge driver fatigue State. When five or more consecutive image displayed that state of the human eye was closed, we determined that driver was fatigue. When five or more consecutive image displayed that state of the human mouth was yawn, we determined that driver was fatigue.