Practical Camera and Colour Calibration for Large Rooms Jacky Baltes Centre for Image Technology and Robotics University of Auckland,Auckland New Zealand j.baltes@auckland.ac.nz http://www.citr.auckland.ac.nz/~jacky Abstract. This paper describes a practical method for calibrating the geometry and colour information for cameras surveying large rooms. To calibrate the geometry, we use a semi-automatic system to assign real world to pixel coordinates. This information is the input to the Tsai cam- era calibration method. Our system uses a two stage process in which easily recognizable objects (squares) are used to sort the individual data points and to find missing objects. Fine object features (corners) are used in a second step to determine the object’s real world coordinates. An empirical evaluation of the system shows that the average and maxi- mum errors are sufficiently small for our domain. Objects are recognized through coloured spots. The colour calibration uses six thresholds (Three colour ranges (Red, Green, and Blue) and three colour differences (Red - Green, Red - Blue, Green - Blue)). This paper describes a fast threshold comparison routine. 1 Introduction Our research work focuses on the design of intelligent agents in highly dynamic environments. As a test-bed, we use the RoboCup domain, which is introduced in section 2. In this domain, small toy cars play a game of soccer. This paper describes an accurate, cheap, portable, and fast camera calibra- tion system (Section 3). After an initial preprocessing step (which is guided by the user), it automatically computes real world coordinates for features in the image (Section 4). Section 5 discusses our algorithm in more detail. The Tsai camera calibration algorithm is briefly described in section 6. Section 7 shows the accuracy that can be obtained by our method in a sample and a real world problem. Both the average and maximum error are sufficiently small for our application. Section 8 discusses the blob detection used in our video server. Objects are identified using coloured spots. The colour detection uses the R-G-B colour model. Each colour is identified by twelve parameters. Six parameters identify the minimum and maximum threshold for the red, green, and blue colour chan- nels. Another six parameters identify minimum and maximum values for the difference channels (red - green, red - blue, and green - blue).