An Adaptive Fuzzy Technique for Real-Time Detection of Multiple Faces against a Complex Background

Abstract

This paper describes a real-time face detection system which is capable of processing video frames extremely rapidly while achieving high detection rate. The primary contribution of this paper is development of a fast algorithm for partitioning each frame into sub-images, detection of potential facial sub-images and real-time clustering of such potential sub-images into isolated objects/faces. A set of experiments in the domain of real-time face detection are presented. The performance of the system is comparable to some well known previous systems [5, 6, 8, 9]. Being implemented on a conventional desktop, face detection could be done at the rate of 13 frames per second.