Eye Tracking System Key Technology Research And Implementation

With the development of eye gaze tracking technology, the technology itself hasmore wider applications in different fields.Eye tracker,as the carrier of eye gazetracking,is always the research focus and achieve difficulty for how to support headnatural movement when it is used.This thesis first introduces the development,classification and application of eyegaze tracking,and describes the eye gaze estimation detailly,which is the key of eyegaze tracking.Then experiment data shows us the influence for tracking accuracybecause of head movement in eye tracker,and some methods to solve head movementin eye tracker are proposed.Next the character input system based on eye gazetracking is introduces,and so is the coplanar sign points method,which is to sovle headmovement of this system.The method solves the mapping relationship between twocoordinate systems in real time and converts coordinates.It consists of scene imageprocessing and projective coordinate transformation.The scene image processingextracts the sign points with the contours detection algorithm based on Hu momentsmatching.The projective coordinate transformation solves projective matrix betweenthe two plane coordinate system.The thesis finally discusses the realization of coplanar sign points method.Itmainly includes the implementation of the scenes image processing algorithm withOpenCV and the scheduling of the overall process by image acquisition cardprogram,which achieves the coordinate transformation of fixation point coordinatesbetween the scene camera coordinate system and the computer screen coordinatesystem.Through experiments,we can achieve the conclusion that the scene imageprocessing algorithm has high extraction accuracy; the range of headmovement iseach side(left and right)15cm,forward5cm and backward10cm. In this range, thetracking accuracy of this system can be basically maintained at less than1°.