Measuring sample distortions in face recognition

In this paper we describe FACE (Face Analysis for Commercial Entities), a framework for face recognition, and show how the approach is made robust to both pose and light variations, thanks to suitable correction strategies. Furthermore, two separate indices are devised for the quantitative assessment of these two kinds of distortions, which allow evaluating the quality of the sample at hand before submitting it to the classifier. Moreover, FACE implements two reliability margins, which, differently from the preceding two, estimate the 'acceptability' of the single response from the classifier. Experimental results show that the overall FACE implementation is able to provide an accuracy (in terms of Recognition Rate) which is better, in some respect, than the present state of art.