The Eye-mouse Research And Implementation Based On Eye Gaze Tracking Technology

Computer vision is the science of how to make the machine to see, further said that means it use cameras and computers to identify, track and measure objects as human eyes. With the popularity of computers and electronic products and their performance being improved, more and more people dedicated to research computer vision.The technology of eye gaze tracking based on eye detection has been widely applied to the driver fatigue detection, disabled living and learning interacting, visual games and other fields.The article describes the principles and methods related to the human eye gaze tracking in the sequence of video images, and researches deep in dynamic face tracking and eye gaze calibration, then proposes the human eye gaze tracking algorithm more suitable for the video image, and combines with OpenCV codes to build a simple human-computer interaction systems making eye gaze as input in VC + +6.0 platform.For face and eye detection, we use machine learning methods based on support vector machine. In order to improve the speed of face tracking, introduce Camshift algorithm based on skin color model to track face in follow-up sequence, because Camshift algorithm is easily affected by the area of skin color, the paper proposed an improved algorithm that is to combine of Camshift algorithm and template matching algorithm to eliminate the area of skin color interference. Experiment shows that the improved algorithm is better than the original algorithm in real-time and immunity, and it can accurately track the human face in complex scenes.After eye is detected, we use circle detection based on Hough transform to locate iris in eye, and get the iris center position in the video window. To get the calibration between the iris center coordinates and gaze position, the article proposed gaze calibration algorithm based on support vector regression, then because it didn't use the traditional polynomial calibration, it improve the accuracy of the eye calibration.The interactive system based on the method developed in the article can real-time track eye gaze, and make the human and computers interact. The system doesn't need many hardware, and can adapt to a certain degree of light changing and head moving such as rotation and pitch, so it is good robust. The method based on the development of this interactive system to achieve real-time eye gaze tracking, and use information detected by the human-computer interaction view. The system requirements on the hardware is not too high, and it can adapt to a certain degree of illumination changes and head slightly change the rotation and pitch, so it is well robustness. The system has the advantage of the characteristics of a cost-effective compared to the current dominant system.