Eye movement tracking calibration method based on head-mounted eye movement module

Abstract

Disclosed is an eye movement tracking calibration method based on a head-mounted eye movement module. A head-mounted module with two cameras is used, and the module is automatically calibrated through a calibration video played by a certain user watching a screen, so that the design purpose of reducing the requirements for the user and the complexity of the calibration process and improving the calibration efficiency are achieved. The method comprises the following steps: (1) enabling a user to wear an eye movement module on the head, and the eye movement module comprises two cameras which are respectively an eye camera for shooting the eyes of the user and a world camera for shooting a display screen; (2) playing a calibration video containing a moving object on a screen, wherein a userwatches the calibration video; (3) obtaining a saliency map set corresponding to the calibration video by using a video saliency detection method, and calculating the position of the moving object inthe screen according to the saliency map set and the position of the video window; (4) calculating user gaze vectors corresponding to the multiple frames of eye images by using an uncalibrated eye movement tracking algorithm; and (5) calculating a mapping relationship between the gaze vector and the screen in combination with the multiple frames of calibration videos and the gaze vector and the screen image corresponding to the multiple frames of calibration videos so as to calculate a screen gaze point corresponding to the gaze vector, thereby realizing automatic calibration.

Description

technical field [0001] The invention relates to a head-mounted eye movement module and an automatic calibration method for eye movement tracking algorithms based on the module, belonging to the fields of computer vision and computer graphics. Background technique [0002] With the rapid development of eye tracking technology, the attention of eye tracking in the field of computer vision is also increasing. At present, most head-mounted eye-tracking modules need to be calibrated for each user before use. There are two reasons for the calibration: First, most head-mounted eye movement modules can only estimate the user's gaze vector, but using the eye movement module actually needs to calculate the gaze point. To calculate the gaze point, you need to know the gaze vector to The mapping relationship of the screen, the mapping relationship is calculated by the calibration process; secondly, the current mainstream head-mounted eye movement module can only estimate the optical ax...

AI Technical Summary

Problems solved by technology

[0003] The existing mainstream eye tracking modules require the user to stare at multiple calibration points for user calibration. This calibration method requires the user to concentrate on looking at the points on the screen. The process is cumbersome and has high requirements for users.

Images