Intrusion Detection System using Contour-based Multi-Planar Visual Hull Method Satoshi KAWABATA 1 , Shinsaku HIURA 1 and Seiji INOKUCHI 2 1 Graduate School of Engineering Science, Osaka University {kawabata, shinsaku}@sens.sys.es.osaka-u.ac.jp 2 Dept. of Kansei Information, Hiroshima International University inokuchi@ieee.org Abstract: In this paper we propose a practical intrusion detection system based on a contour-based multi-planar visual hull method. The system consists of several sets of a camera and a PC connected via network. Prior to the monitoring, a system mamager will set up several sensitive planes in the world using a GUI. To detect the intruding objects, images taken by the cameras are compared with a maintained background image using PISC method. Then the contour of the object region on a image is traced and transformed to the coordinate of sensitive plane. The contours are collected to a single PC and existence of a common region is judged. If the intruding object exceeds the sensitive plane, the contours from each PC would have some common regions. If the intruder occludes the critical point on the sensitive planes from the all cameras to avoid observation, the system also alerts to the user. This system does not reconstruct the shape of the object, but the system is able to handle 3-D protection volumes because multiple sensitive planes can be set arbitrary. This system does not require us to calibrate the camera because this system is based on 2-D homography. Keywords intrusion detection system, visual hull, PC cluster, multisensor 1 Introduction Nowadays a lot of cameras are used at shops, museums, banks, roads and so on, but the images are mainly used for evidence for postmortem. In the case of very criti- cal missions which requires real-time responses, human would watch many monitors on a wall. To save the la- bor and cost for monitoring, special sensors are often used. Light beam detectors are used to block the intru- sion, but the sensitive lines are not flexible because it is strictly related to the arrangement of the equipments. Passive infra-red sensors which detects the body heat is also widely used, but it will detect only the existence of the human body and detailed coordination of mon- itoring area is impossible. For replace human observer with a computer, image-based surveillance and moni- toring method is extensively researched. For example, the DARPA VSAM project[1] and ISPS CDV project[2] are intensively attacked to the problem and produced many successful results. However, most methods pro- posed by these projects are complicated and difficult to use because they tried to understand the action of the object. Contrary to these projects, we aimed to only de- tect the intruder who goes across critical surfaces which is arranged by a supervisor. It is very simple but generic and easy to understand, and we believe that it is enough satisfactory to the most applications. The important properties of actual monitoring system are as follows: 1) easy to install, 2) robustness, 3) responsibility to the unexpected events. In fact, our system does not restrict the shape of the critical surfaces (sensitive planes) and it is not related to the location of the cameras. The cameras are not necessary to be calibrated in advance, and the sensitive planes can be set on the input image displayed on a GUI interface. Since our system is based on the background subtraction method, we employed robust PISC algorithm. It is robust to the change of illumination or shadow of the object. Our system will alert when the critical point (intersection of the object and sensitive planes) is occluded by the other object. This characteristic is preferable to the monitoring sys- tem. 2 Overview of Intrusion Detection System 2.1 System Construction For our monitoring system, we use four cameras at each top corner of cubic shaped space as shown in figure 1. However, the number and arrangement of cameras is not a matter of the shape of the monitored space. At least two cameras are necessary to apply our monitoring