Research And Implementation Of Driver Fatigue Detection Based On Digital Image Processing Technology

In our country, as people’s living standards improve, the amount of cars grows rapidly. But more and more traffic accidents come with it, bringing huge damages to the society. Fatigue driving is one of the main factors adducing traffic accidents. It is an effective means to avoid these accidents by monitoring the driver’s fatigue status and warning timely. So it has practical significance to study and implement fatigue detection algorithm.On the basis of researching the current technology of fatigue detection both at home and abroad, this paper presents the overall design of the driver fatigue detection scheme and makes a simulation in actual scenario. In the positioning stage, the Adaboost cascade classifier was used to detect the target. The human face was detected first, and then search the human eye within the face scope. Thus, the speed of detecting human eye was improved. The driver’s eyes which were to be picked as eye templates were selected from the detecting result. Template matching method was used to track the human eye in the tracking phase. When the tracking was lost, reposition was to find the eyes. Predicting the human eye area before template matching can reduce the search scope and improve the accuracy and speed. With the thought of uniformly variable, this paper predicted the area through the displacement of the three frames before. In order to solve the eye tracking lost problem of traditional template matching, which were due to the accumulated error and human eye blinking, updating template was introduced in this subject. In the fatigue evaluation stage, eye blink frequency and PERCLOS are obtained from human eye state parameters. This paper presents a new fatigue testing process to use the two indicators of fatigue more rationally. That is using different indicators in different fatigue stages to make it more practical.The fatigue detection algorithm simulation of this subject is in VS2008integrated development environment, using C#programming language and called OpenCV. The test result under different light, movement and rotation angles shows that:in normal sunlight, the driver’s fatigue detection scheme of this paper can be precisely implemented on time in every stage, and can reflect the driver’s fatigue extent.