视线跟踪中基于虹膜识别的一次标定方法

One-Time Calibration Method Based on Iris Recognition

为了简化头戴式视线跟踪系统中的标定过程,提出一种无需添加任何硬件的、基于虹膜识别的一次标定方法.使用者只需在第一次使用头戴式视线跟踪系统时进行标定,再次使用时,系统会自动进行虹膜识别,调出使用者第一次标定时的眼部图像数据来计算当前眼部图像与标定时的眼部图像的相对偏转角和偏移量,得到使用者当前的标定参数.实验结果表明,该方法可以在不影响系统原有精度的条件下,省去使用者每次使用视线跟踪系统时都需要进行的烦琐的标定过程,大大简化了系统的标定,降低了系统的使用复杂度.

With the aim to simplify the calibration process in the head-mounted gaze tracking system, a one-time calibration method was proposed, which was built upon the iris recognition system and did not require any extra hardware. At the first time when the user took on a head-mounted gaze tracking system, the camera calibration was conducted. The eye image used for the calibration was recorded with the corresponding user＇s ID information. In the subsequent usages, the system automatically recognized the user by the iris recognition. The eye position were extracted and compared with the one captured at the calibration stage. The new calibration parameters were able to estimate based on the relative rotation angle and offset between these two images. The experimental results demonstrated that this method was able to avoid conducting the cumbersome calibration process each time when the relative position of an eye and the camera was changed but without affecting system＇s performance. So it greatly simplified the calibration process, and reduced the complexity of the usage of the head- mounted gaze tracking system.