A video-based human eye position detection method

Abstract

The invention discloses a video-based human eye position detecting method. According to the video-based human eye position detecting method, positive and negative extremum pixel points which are formed through the reflection of a light source at human eyes are adopted as the characteristic points of a human eye region, and therefore, the method has excellent pseudo-extremum point suppressing ability, is small in computation burden and can realize fast detection. With the human eye position detecting method provided by the invention adopted, human eye detection can be carried out at regions except paraxial regions of a camera, and requirements for the positions of the human eyes and the camera are lowered. The video-based human eye position detecting method can carry out detection in a large view field range.

Description

Technical field [0001] The invention belongs to the field of image processing and machine vision, and relates to a method for detecting the position of human eyes based on video. Background technique [0002] The position information of human eyes is of great significance for face positioning and gaze tracking. Currently, there are two types of methods for detecting the position of the human eye: an active detection method that uses an external light source to illuminate the pupil and a passive detection method that does not use a light source to illuminate the pupil. Passive detection methods mainly use machine learning methods, which require a large number of samples to train the classifier to have a better detection result, and the training process takes a long time. The traditional active detection method uses the bright-dark pupil difference technology. The realization of this method needs to ensure that the human eye position is always in the paraxial region of the imaging...

AI Technical Summary

Problems solved by technology

The passive detection method mainly adopts the machine learning method, which requires a large number of sample training for the classifier to have a better detection result, and the training process takes a long time

The traditional active detection method uses the bright and dark pupil difference technology. To realize this method, it is necessary to ensure that the position of the human eye is always in the paraxial region of the imaging system during the detection project. This method is very sensitive to the position of the human eye and the camera, and the working range is limited. In addition, after the difference, various forms of filtering and denoising processing are required to accurately identify the position of the human eye. The amount of calculation is large, and it is difficult to achieve fast detection.

Images