Application Of Gaze-tracking In The Wheelchair Control System

With the rapid development of computer vision technology and the rise of artificial intelligence research,gaze-tracking technology has become a research hotspot in major research institutions and universities.As a new Man-machine interaction technology,gaze-tracking is widely used,including intelligent medical treatment,assisted driving,psychological research,virtual reality,military and other fields.The extremely rich human eye movement information can reflect the changes of human emotions and behavioral trends,and is also the most intuitive embodiment of visual tasks.Therefore,when selecting and manipulating visual objects,some channels of Man-machine interaction can be conducted under the guidance of visual channels,that is,users watch the objects they're interested in first,and then guide the visual channels to select or manipulate actions.To sum up,using eye control system can manipulate visual objects more conveniently and promptly.Focusing on the subject of gaze-tracking,the thesis used the line-of-sight estimation method based on pupil-Corneal reflex to construct the head-mounted gazetracking control wheelchair system.Compared with the non-head-mounted gaze tracking control wheelchair system,this system has the advantages of high precision and wide application range.The methods of pupil location,the calculation of the point of regard and the eye control wheelchair scheme are discussed:1?The extraction algorithm of pupil and Pulchin Spot Center is studied.The ellipse fitting algorithm based on arc segment extraction is used to locate the pupil center.In order to solve the non-linear non-Gaussian problem and improve the extraction accuracy of pupil state parameters,this paper uses the combined method of particle filter and Kalman filter to track the pupil,which significantly improves the robustness of the system.Then,the area around the pupil is intercepted,and the image is binarized and segmented into the Pulchin Spot by using OSTU adaptive threshold method.Finally,the center of the Pulchin Spot is calculated by using the centroid method.2?The principle and process of the calculation of the point of regard are analyzed.The nonlinear mapping relationship between the eye image coordinate system and the scene image coordinate system is calculated by using the real point of regard impact point coordinates and eye movement parameters,and the viewpoint coordinates betting on the scene image coordinate system are estimated.In this paper,BP neural network is used to fit the mapping relationship,and the given 15x10 samples are used to train the neural network to determine the mapping relationship between the eye image coordinate system and the scene image coordinate system.3?The eye-controlled wheelchair system is constructed.By calculating the relative position between the central coordinate of the pupil and the reference point,the relative position is converted into a direction command,which is sent to the wheelchair controller through Bluetooth,and finally,the motion direction of the wheelchair is controlled.The experimental results show that the accuracy rate of the point of regard estimation is 91%,when the average error of gaze-tracking is less than 5 units,the average accuracy rate of moving five ways of the wheelchair is 98.2%,which are forward,backward,left,right and stop respectively.The movement of wheelchairs can be smoothly controlled through eye movement parameters,thus providing new ways for the disabled,amyotrophic lateral sclerosis and the elderly to move freely.