Automatic Human Faces Detection and Localization using Skin Colour

Time: 11am Friday 2 April

Computer Science & Software Engineering

Seminar Room 1.24

SPEAKER: A. (Salim) Bouzerdoum

Abstract:
The problem of automatic detection and localization of human faces in digital images and video is a crucial first step in many applications that involve facial image analysis. For example, in a face recognition system, before a face can be compared to stored faces in a database it must be extracted automatically from a visual scene. Robust real-time detection of faces is needed in a video surveillance system where dynamic scenes are continually scanned for known faces. Furthermore, face identification and tracking is required by service robots operating in a natural environment for identification and interaction with humans. In all these applications, accurate and fast human face detection is the key to a successful operation.

Our aim is to develop an advanced automatic vision system for detecting, locating and tracking human faces in highly complex and dynamic environments. Both analytic and holistic face detection techniques will be combined to develop novel and efficient methods that are invariant to rotation, scale, lighting conditions, and facial expressions. The proposed approach relies on skin colour information and other facial features to reduce the search space and accurately locate the position and orientation of the face in the image. The system consists of four major components: skin segmentation, face candidate selection, face verification, and face tracking. In this seminar, we will report on the progress to date achieved in the first three components. To this end, we have developed a number of classification methods for skin colour and face patterns, based on Bayesian decision theory and neural networks. A rigorous evaluation and comparison of the performances of the different classifiers will be reported.

Short Biography.
Dr. Bouzerdoum received the Masters and Ph.D. degrees, both in electrical engineering, from the University of Washington, Seattle, in 1986 and 1991, respectively. From 1991 to 1997 he was with the University of Adelaide, South Australia. In 1998 he was appointed Associate Professor at Edith Cowan University. Dr. Bouzerdoum has received several fellowships and distinguished awards; amongst them are the Vice Chancellor’s Distinguished Researcher Award in 1998 and 1999, and the Awards for Excellence in Research Leadership and Excellence in Postgraduate Supervision. In 2001 he was the recipient of a Distinguished Researcher (Chercheur de Haut Niveau) Fellowship from the French Ministry of Research to spend three months at the National Research Centre LAAS - CNRS in Toulouse, and in 2003 He was awarded a Visiting Professor position from the University of Paris-13. He is currently serving as Associate Editor for the IEEE Transactions on Systems, Man and Cybernetics, and Chair of the IEEE Signal Processing Chapter of WA Section. He is a senior member of IEEE, a member of INNS, and a member of IASTED Technical Committee on Neural Networks. He has published over 180 technical articles and graduated 12 Ph.D. and 4 Master students. His research interests include signal/image processing, pattern recognition, machine vision and machine learning, and VLSI implementation of smart vision micro-sensors.