I. Introduction
In state-of-the-art applications of intelligent surveillance systems (ISS) the main goal is to make them more operator- independent and also to make the decision process in these systems more similar to human decision making procedure. Simultaneously the developments in the field of image processing, pattern recognition and computer vision make the algorithms more reliable, faster and more robust than before. Consequently the developments in those areas provide suitable tools for surveillance system designers to approach the abovementioned goals.