Towards Detection of Glasses in Facial Images

Abstract

In this paper we consider the automatic detection of the presence of glasses. Using the eye locations detected in facial images, two regions of interest around the eyes are investigated where glasses are expected. We introduce six measures for the likelihood of glasses. An experimental evaluation on two image sets demonstrates reasonable results of separating facial images with and without glasses. Furthermore, a combination of measures turns out to consistently improve the performance of the individual measures. We also perform a sensitivity analysis by means of simulation to figure out the inaccuracy of eye detection that can be tolerated by our glasses detection method.