A Gaze Tracking Method Based on Neural Network Head Motion Compensation

Abstract

The invention discloses a sight tracking method based on neural network head movement compensation, and the method comprises the following steps of respectively placing two infrared sources on middle points of two side edges of a display screen, forming a bright spot image on the surfaces of human corneas after the two infrared sources are reflected by the human corneas, and enabling a dark pupil effect to occur in pupil images, and the method specifically comprises the following steps of statically calibrating a fixation point; compensating head movement; and calculating the fixation point by utilizing a static mapping module and a head movement compensation module to form a dynamic human eye fixation point matching module. Due to adoption of the scheme provided by the invention, requirements on hardware are low, only an ordinary camera and two infrared sources are needed, the restriction of a three-dimensional tracking method in which multiple cameras are needed can be changed, the system complexity and the needed computation volume can be greatly reduced, and the sight tracking instantaneity can be improved; and in addition, the head movement is allowed, direct application can be realized through one-step calibration, repeated calibration is avoided, and main defects of the two-dimensional calibration method are overcome.

Description

technical field [0001] The invention belongs to the technical field of image processing and pattern recognition, and in particular relates to a line-of-sight tracking method based on neural network head motion compensation. Background technique [0002] Human information processing relies heavily on vision, and more than 80% of the information from the outside world is obtained through the eyes. Therefore, the human perception and intention of external information can be obtained by tracking the gaze point of the human eye, and the It is used in various fields and plays an active and effective role. For example, eye-tracking technology has broad application prospects in the fields of new-generation human-computer interaction, military armament development, automobile and aircraft driving, aerospace, sports, assistance for the disabled, virtual reality and games, and has a great impact on improving the lives of disabled people. It is of great practical significance to build a...

AI Technical Summary

Problems solved by technology

Therefore, the 3D eye-tracking technology has defects such as complex structure, difficult modeling, and poor real-time performance.

In contrast, the 2-dimensional tracking method locates the eye gaze point by determining the mapping function, which is easier to implement and more widely used than the 3-dimensional tracking method, but it also has two main disadvantages: 1. It is sensitive to head movement. When the head moves, the positioning accuracy will drop rapidly; 2, the calibration process must be carried out before each use, so it is also very inconvenient, which is not conducive to wider application in practice

Images