Design And Implementation Of Human Computer Interaction System Based On Gaze Tracking Technology

The gaze tracking technology is a hot topic in the field of computer vision and the human-computer interaction based on gaze tracking technology has been attracting more and more researchers' interest in the field of multi-channel natural interaction.The application of gaze tracking technology for human-computer interaction can change the traditional human-computer interaction mode,which can bring new,natural and efficient interactive experience to users and help the physically disabled people communicate with computers naturally as normal.so the subject has a certain academic value and social significance.This thesis studies typical gaze tracking technology,and designs gaze tracking technology based human-computer interaction system combined with stereo vision technology.The main work of hardware design includes the following aspects.Designing of the image acquisition module to capture the user's face image.Designing of the cursor control module to interacte with the computer.Designing image output module to complete the real-time video image display.By using the SoC technology,it integrats the ARM hardcore processor,image acquisition module,the image output module and the cursor control module and finally implements the overall hardware design.The software design of embedded system implements as following aspects.Transplanting embedded Linux operating system on the platform to achieve multi-tasking.Transplanting Qt graphics library to design display interface.Transplanting OpenCV computer vision library to help achieving part of the image processing algorithms.Developing the upper application,detecting the characteristics related to the direction of the gaze,calculating the user's face orientation,completing the positioning of the user's gaze,and finally controlling the computer cursor to achieve human-computer interaction.The contributions of this thesis are as follows.First the method of gaze tracking through calculating the face orientation with stereo vision is proposed,which allows the user's head move within a certain range and tracks the user's gaze accurately.Second the Iris localization algorithm based on circle operator and boundary method is designed for solving the problem of iris location in low-resolution human eye image under natural light.Last using the SoC technology can enhance the scalability and benefit the following function upgrade and rapid commercialization.The test results show that the system can achieve the gaze tracking when user view the monitor,and can control computer's cursor based on the gaze tracking results correspondingly,at last can achieve human-computer interaction through the line of sight.