Image Preprocessing for Eye Pupil Tracking

摘要

The influence of image preprocessing methods on the quality of eye pupil tracking by video is investigated. The methods of blurring, sharping, contrasting and scaling are considered. An estimate of the signal-to-noise ratio for image recognition based on the signal spectrum is proposed. The results obtained make it possible to determine combinations of image preprocessing modes that give the best pupil detection accuracy and a sufficiently high frame processing speed.