Drivers' eyes open or closed state of the computer image recognition technology development
|Keywords||Human eye recognition Infrared differential Template matching Kalman filtering|
The eyes are the most important basis for judging the driver fatigue through the identification and tracking of the driver's eyes open or closed state can be more accurately determine whether the driver fatigue , while real-time high , and easy to implement . Although driver fatigue detection system is to detect the multi-parameter , multi - information fusion system development , but changes in the eye in numerous recognition parameters always occupy the most important position , and therefore improve the eye recognition accuracy rate and fast tracking eye fatigue development the driving detection system has a significant role. Combined research background and research status , this paper, identification and tracking algorithm based on infrared pupil do bad eyes , and build a build based on the DM642 embedded hardware platforms and software platform based on ICETEK-DM642-BR identification system of the human eye , as follows : ( 1) production of controllable infrared light source image acquisition board . First selected 850nnm infrared LED as a lighting source , followed by the design of the LED drive circuit , and the drive circuit of the various components were selected , the light source through the DM642-BR board GPIO expansion interface control . (2) The combination of infrared difference image and computer vision technology pupil localization algorithm . The brightness pupil images obtained in different infrared differential, the differential Gaussian smoothing filter is used after the image processing and after experimental method was adopted to select a suitable threshold for image segmentation , and finally extract the location of the pupil . (3 ) improved template matching algorithm to directly determine the status of the opening and closing of the eyes , compared to the traditional template matching algorithm does not need to extract the characteristics of the eye , can save time , improve recognition speed . ( 4) the use of the Kalman filter to track the eyes . Kalman filter predict increased the speed of the identification and tracking of eyes . (5 ) structures of the eye recognition algorithm based on the the DM642-BR board hardware platform and software platform , and through the CCS test and optimize it to meet the requirements of real-time , accurate .