Driver's Eyes Opening And Closing State Computer Image Recognition Technology Development

Eyes are the most important factor that judges whether the driver is fatigue. We can judge whether the driver is fatigue by the recognition of the open and close condition of the eyes, the method can meet the real-time and is much easier to achieve. Despite the current fatigue driving detection system develops pluralistic, the changes in the eyes occupy a most important position in so many recognition parameters. Therefore improving eye identification accuracy and quickly tracking eyes have the significant role in developing fatigue driving detection system.Given the background of the research and the current studies from home and abroad, this thesis designed a method that identify and track eyes based on difference picture IR and image processing. Meanwhile we built the hardware platform and software platform based on ICETEK-DM642-BR and construct the embedded system of identifying eyes. The main work is as follows:(1) Make an image acquisition board that can control the infrared light source. First select 850nm infrared LED as the lighting source, then design the LED driving circuit and select the components on the circuit. The light can be controlled through the GPIO outside enlarge interface on the DM642-BR.(2) Propose an algorithm of pupil localization which is combined by infrared difference image and Computer Vision Technology. The algorithm does differential treatment to the bright or dark pupil image got from different infrared light, then smooth the image with Gauss filter, and then select proper threshold to partition the image, finally extract the pupil location.(3) Put forward an improved template matching algorithm to directly estimate the condition of eyes. Comparing to the conventional algorithms, this method doesn't need to extract the feature of eyes, so it can save time and improve the speed of recognition.(4) Make use of the Kalman wave filter to track the extracted features. The speed of recognition and tracking of eyes can be accelerated with the prediction function of Kalman wave filter.(5) Construct hardware platform and software platform based on DM642-BR board. Test and optimize various eyes detecting algorithms in order to meet the requirements of real-time and accuracy.