Combined Approach for Face Frontal View Estimation for Real Time Faces Detection in Мulticamera System

Abstract

This paper presents the combined approach for face frontal view estimation from video sequences in a multiview camera setup. This task is important for person identification by face image in video surveillance systems. Face tracking algorithm was developed based on optical flow and cascade face detector. We also found way to estimate quality of face detection. This quality is used as base for best frontal view estimation.