@TechReport{TR041,
  author =	 "Thomas Nemec",
  institution =	 "PRIP, TU Wien",
  number =	 "PRIP-TR-041",
  title =	 "Gesichtserkennungs-{S}ystem zur {R}aumueberwachung",
  year =	 "1995",
  url =		 "https://www.prip.tuwien.ac.at/pripfiles/trs/tr41.pdf",
  abstract =	 "The goal of this diploma thesis is to develop a face
                  recognition system that uses CCD camera fixed on a
                  robot arm. In order to extend the ``visual field''
                  of the system the robot moves with the camera. The
                  task of the system is to find people and to identify
                  them. After a general introduction some methods of
                  face recognition and detection are explained. The
                  structure of the system and its tasks are
                  presented. To evaluate the whole system a database
                  consisting of people from the staff of the
                  department ist created. The goal is then to check if
                  a person entering the room belongs to the staff or
                  not. In case the person belongs to the staff its
                  name should be given. \\The system consists of the
                  following major modules: Motion detection, Face
                  detection, and Face recognition. In particular, the
                  following methods are used within the modules:
                  Motion Energy detection for Motion detection,
                  Multilayer Perceptrons for Face detection, and the
                  Eigenface approach of Turk \& Pentland for Face
                  recognition. \\After evaluating the individual
                  modules we tested the face recognition system on
                  different people entering the observation
                  room. Provided one person stays app. 10 seconds in
                  the observation room, the overall rate is 82,4 \%
                  which is a good result considering the speed of one
                  frame per second on fully automatic face
                  recognition.",
}