Design And Implementation Of Visual Control Software Platform Based On OpenCV

Eyes are the main channel for humans to obtain information from the outside world.It has become a hot topic to determine the direction of sight through the shape and characteristics of human eyes.Gaze tracking is a technology that uses the movement of the human eyes to estimate the direction of the line of sight.With the use of computers,gaze tracking technology has been used in fatigue driving,medical diagnosis,psychology and military fields.Based on the research of existing gaze tracking algorithms,this thesis proposes a pupil center positioning algorithm based on the principle of concentric circles of pupils and iris for pupil positioning and the single-point calibration algorithm based on coordinates to determine the eye direction.Based on this algorithm,a visual control software platform based on Open CV is designed and developed,which mainly designs the algorithms and platform software required for the entire platform operation.In terms of gaze tracking algorithm design,firstly,skin color detection is used to determine the general position of the face,and the image detection range is reduced.Then the detected image is pre-processed,include image graying,bilinear interpolation,and histogram equalization for face detection and localization.Afterwards the eyes are detected and located by using the proportion of "three courts and five eyes".Then the image gradient algorithm is used to estimate the pupil center,then the iris center is determined by the method of fitting based on the improved least square method,and the evaluation standard of fitting rate is made according to the estimated pupil center and the fitted iris center.Finally,the single-point calibration algorithm based on coordinates is used to determine the line of sight.The visual control software platform uses Visual studio 2017 as the development tool,combines the computer vision library Open CV3.4.1 with the Microsoft Foundation Classes(MFC),and uses C++ language to program algorithms into the platform software.In the image interface design,modules are created,including face detection and positioning,human eye detection and positioning,pupil center positioning,and gaze direction determination are created.A simple application of visual control for playing piano is developed.Through the experimental test of the algorithm of the whole platform,the visual control software platform based on Open CV designed in this paper can achieve real-time and accurate line of sight tracking,with an accuracy of 96.25%,and can overcome the problems of head movement,glasses,eyelids and other errors brought by the detection,with good robustness and accuracy.