Sight line detection and video processing method and device, equipment and storage medium

Abstract

The embodiment of the invention discloses a sight line detection and video processing method and device, equipment and a storage medium. The sight line detection method comprises the steps that according to key feature points in a human face image, a corresponding human face posture and eye pupil rotation displacement are determined, and the eye pupil rotation displacement is the displacement of the pupil center relative to the eyeball center in the human face image; and according to a preset projection function and the human face posture, the eye pupil rotation displacement is projected to athree-dimensional space where an actual human face is located to obtain a sight line direction of the actual human face. According to the technical scheme provided by the embodiment of the invention,the processing amount of the eye data is reduced, the detection efficiency of the sight line direction is improved, the sight line direction of the actual face is analyzed directly through eye pupil rotation in the face image, and the detection accuracy of the sight line direction is improved.

Description

technical field [0001] Embodiments of the present invention relate to the field of image processing, and in particular, to a method, device, device, and storage medium for line-of-sight detection and video processing. Background technique [0002] With the development of video technology, 3D virtual characters have been widely used in entertainment, movies, and virtual reality (VR) simulations. At this time, eyes are a very important part of virtual characters. There are much fewer body movements, but people are very sensitive to eye movement perception, unnatural eye angles and movements are easily noticed, and the eye movement capture device that collects the center point of the pupil of the eye is usually worn on the head It is very inconvenient to use. [0003] In response to the above problems, in the application scenario of realistic facial expression animation, the transformation from the captured pupil center point to eyeball movement is usually realized by using th...

AI Technical Summary

Problems solved by technology

[0009] Among them, literature [1] and literature [2] are a similar data-driven method, and literature [3] uses a simpler and more intuitive bulletin board method, which is to select from a large number of eyeball textures that best match the current eyeball state. The eyeball texture is placed on the target eyeball to realize the migration of eyeball movement. It needs to be compared with a large number of historical eyeball textures. The amount of data processing is large, resulting in low efficiency of sight judgment; literature [4] directly uses the movement of eye muscles to To track eye movement, literature [5] adopts a rule-based method to directly synthesize various movement effects of line of sight direction changes, but does not directly analyze eye movement according to pupil position changes, making the accuracy of the determined eye line of sight low

Images