IMAGE-BASED 3D ACQUISITION TOOL FOR ARCHITECTURAL CONSERVATION Joris Schouteden, Marc Pollefeys, Maarten Vergauwen, Luc Van Gool Center for Processing of Speech and Images, K.U.Leuven, Kasteelpark Arenberg 10, B-3001 Leuven, Belgium Tel. +32 16 321064; Fax +32 16 321723 Marc.Pollefeys@esat.kuleuven.ac.be KEY WORDS: photogrammetry, architecture, heritage conservation, image-based 3D reconstruction, dense stereo. ABSTRACT In this paper an image-based 3D acquisition tool is presented. The tool has been developed with architectural conservation in mind. The proposed tool is developed as part of a project that aims at developing a technology that enables the operator to build up an accurate three dimensional model - without too much repetitive work - starting from photos of the objects and measured reference co- ordinates. This model can in a later phase be used as a core for a multimedia database, to represent designs for interventions, or for distribution to the wider public. The presented 3D acquisition tool is largely automated. A limited amount of interaction is used to indicate the reference points in the images to allow global localization in world coordinates. These points are also used to simplify the matching process so that wide baselines can be coped with. Depending on the number of reference points the camera needs to be pre-calibrated or not. From this point on the approach automatically finds additional matches and refines the calibration. A globally consistent estimate can be obtained through bundle adjustment. The next step consists of dense depth estimation based on stereo matching. From the computed information a dense textured 3D surface reconstruction of the recorded structure is computed. The tool also allows extracting 2D slices or pointing measurements. The results can be used both for measurements and for visualization purposes. 1 INTRODUCTION At this moment many architects involved in conservation still work in the traditional way. They use instrument based (theodolite, total station, photogrammetry) or even hand-measured (tapes, plumb-bobs, levels...) survey methods. As many architects are shifting towards computer-aided design for new buildings, they also try to apply these programs to renovation or conservation projects. However, the number of tools available to accomplish the task of 'getting the existing building in the CAD program' is limited, and mainly directed to 'translate' traditional methods to CAD (automatic import of full station co-ordinates, error-adjustment of triangulation...). Based on a limited number of actually measured points, 2D plans and sections or a 3D model can be constructed. This typically results in a very 'simplified' representation of the building, which is absolutely not in line with the high requirements for conservation purposes. The presented 3D acquisition tool is largely automated. A limited amount of interaction is used to indicate the reference points in the images to allow global localization in world coordinates. These points are also used to simplify the matching process so that wide baselines can be coped with. Depending on the number of reference points the camera needs to be pre-calibrated or not. From this point on the approach automatically finds additional matches and refines the calibration. A globally consistent estimate can be obtained through bundle adjustment. The next step consists of dense depth estimation based on stereo matching. A complete model is obtained by combining the results of different disparity maps into a 3D reconstruction. The texture is extracted from the original photographs. From the computed information a dense textured 3D surface reconstruction of the recorded structure is computed. The tool also allows extracting 2D slices or pointing measurements. The results can be used both for measurements and for visualization purposes. The proposed tool is developed as part of a project that aims at developing a technology that enables the operator to build up an accurate three dimensional model - without too much repetitive work - starting from photos of the objects and measured reference co- ordinates. This model can in a later phase be used as a core for a multimedia database, to represent designs for interventions, or for distribution to the wider public (Nuyts 2001). 2 CALIBRATION The first module allows to calibrate the camera and to compute the pose in world coordinates for the different views. For this purpose the tool assumes the knowledge of 3d reference points measured by surveying methods as well as their projection in the imagery. As a part of this tool an interface is provided for identifying the 3d reference points in the images. A snapshot of this interface can be seen in Fig. 1. The reference points have been indicated in both images.