In this paper, we present a complete system for the analysis of architectural drawings, with the aim of reconstructing in 3D the represented buildings. We successively describe the graphics recognition algorithms used for image processing and feature extraction, the 2D modeling step, which includes symbol recognition and converts the drawing into a description in terms of basic architectural entities, and a proposed 3D modeling process which matches reconstructed floors. The system also includes a powerful and flexible user interface.

In this paper, we discuss the elements to be taken into account when choosing one's vectorization method. The paper is extensively based on our own implementations and tests, and concentrates on methods designed to have few, if any, parameters. ¢ This work is partly funded by France Telecom CNET. £ Now with Business Objects, Paris.

We claim that time has come in graphics recognition for choosing stable and robust methods, even-or especiallywhen this means implementing methods proposed by others, instead of inventing a new algorithm which ends up being a minor variation on an old idea. In this spirit, we present some of the choices our own team has made.

Abstract. This paper presents the main design and development issues of the Qgar software environment for graphics recognition applications. We aim at providing stable and robust implementations of state-of-theart methods and algorithms, within an intuitive and user-friendly environment. The resulting software system is open, so that our applications can be easily interfaced with other systems, and, conversely, that thirdparty applications can be "plugged" into our environment with little effort. The paper also presents a quick tour of the various components of the Qgar environment, and concentrates on the usefulness of this kind of system for testing and evaluation purposes.