Eye and nostril localization for automatic calibration of facial action recognition system

Abstract

The work presented here focusses on automatic facial action recognition using image analysis algorithms and application of facial gestures to machine control. There are many sources of variation in facial appearance which make recognition a challenging task. Therefore, machine adaptation to human and environment is — in our opinion — the key issue. The main contribution of this paper is eye and nostril localization algorithm designed to initialize facial expression recognition system or recalibrate its parameters during execution.