Face and Pose Recognition for Robotic Surveillance

Face recognition has been a huge area of research over the past 25 years [Gao and Leung, 2002]. However the field is still highly unsolved, largely due to variations in pose, illumination and expression. In this paper we propose using pose recognition to solve the first of the face recognition variation problems and simultaneously use both recognised pose and face information to control a robot in a surveillance application. The main advantage of using face and pose recognition to control a robot is a more natural man/machine interaction with the recognised controller (with sufficient clearance) being able to control the robot from their point of view. Also, less manual (hands-on) controls will be required, if the control system is mainly visual. Finally in crowded environments visual control cues from a recognised person are more robust and secure than audible control cues (which are the alternative to either visual or hands-on cues). The recognition method used for both the face and pose recognition is based on geometric 3-Dimensional feature point matching.