计算机技术与发展2017，Vol.27Issue(1)：39-43,5.DOI:10.3969/j.issn.1673-629X.2017.01.009
基于深度相机的水平视角估计及校准方法
An Approach of Horizontal Gaze Angle Estimation and Calibration Using Depth Camera
摘要
Abstract
An approach to estimate and calibrate the horizontal gaze angle using the Kinect depth camera is proposed. Fisrt of all,the Ki-nect is used to capture the face image and estimate the head pose,localizing the 2D pupil center by means of image gradient. Then the 3D pupil center is obtained by the intrinsic parameters of the Kinect. With the pupil center and an empirical eyeball center value,the horizon-tal gaze angle can be estimated conveniently. Unlike the previous methods which need to calibrate the eyeball center for every person,the method is based on the predefined empirical eyeball center,which the gaze estimation system is robust and convenient. In order to improve the accuracy of the gaze angle,an experimental system is built which can collect the gaze angle and display the gaze target. The several participants’ data of measurement data and the ground truth are collected. A linear regression model is built from the errors between the ground truth and the measured data,and used to correct the accuracy of the estimated gaze direction. The experiment shows that the ap-proach can fast and accurately estimate the gaze angle.关键词
校准方法/水平视线角度估计/Kinect深度相机/线性回归模型/拟合函数Key words
calibration approach/horizontal gaze angle estimation/the Kinect depth camera/linear regression model/fitting function分类
信息技术与安全科学引用本文复制引用
倪剑帆,汪建明,周小芹,刘小峰..基于深度相机的水平视角估计及校准方法[J].计算机技术与发展,2017,27(1):39-43,5.基金项目
中央高校基本科研业务费专项资金(2015B38214) （2015B38214）
常州市特种机器人与智能技术重点实验室开放基金项目(CZSR2014003) （CZSR2014003）
