Virtual Reconstruction of Heritage Sites: Opportunities and Challenges Created by 3D Technologies *

This paper looks at the opportunities and challenges created by new 3D technologies introduced recently in the context of virtual reconstruction of cultural heritage sites for preservation and entertainment. A number of projects carried out by the authors in the last four years support the conclusions made in this paper. The projects presented include a 6th century bronze sculpture, metopes/architectural elements of a Greek temple, and, a Byzantine Crypt. Results on CDROM, DVD, 3D theatres, holograms and, video animations have been prepared for these projects. The 3D modeling work that was accomplished in preparing these rich-media products is based mainly upon high-quality photo-realistic texture mapping onto high-resolution 3D models generated from range images. This procedure is enhanced with the integration of both photogrammetric and CAD modeling techniques. 2 OPTICAL SENSORS FOR SHAPE ACQUISITION When describing and explaining the history of a heritage site, the use of spatial information facilitates the understanding of that particular site. Laser scanners represent an effective way to create high-resolution 3D models of the existing condition of a heritage site or a cultural artifact. Two main classes of laser scanners are reviewed with a particular attention to the fundamental measurement limits. The goal is not to survey all commercial 3D vision systems. Instead, some basic theory about 3D sensing is presented and is accompanied by selected results that should give some pointers in order to become more critical when choosing a 3D solution. 2.1 Range uncertainty of laser scanners for 3D imaging applications Active 3D imaging sensors can be divided using different taxonomies (Jähne et al. 1999). The main classes are − Triangulation: Single spot (1D), profile measurement (2D) coupled to mechanical motion devices, position trackers, mirror-based scanning, Area measurement (3D really 2.5D) based on multi-point and line projection diffraction gratings, fringe pattern projection, Moiré effect, − Time delay systems: Single spot with mirror-based scanning based on pulses, AM or FM modulation, Full field (scannerless) using micro-channel plates or custom build silicon chips, Triangulation fringe projection systems are not considered here. For close range 3D applications, they offer interesting characteristics (Jähne et al. 1999). 2.1.1 Triangulation The basic geometrical principle of optical triangulation is shown in Figure 1a. The measurement of range operates as follows. A light source projects a beam on a surface of interest. The scattered light from that surface is collected from a vantage point distinct from the projected light beam (distance D). This light is focused onto a position sensitive detector (herein called spot sensor). The knowledge of both projection and collection angles (α and β) relative to a baseline (D) determines the dimensions of a triangle. The complete range equations are derived in Blais (2004). For an incremental change of distance, ΔZ, one measures the incremental angle shift Δβ. This laser spot sensor is in fact an angle sensor. The angular shift Δβ caused by a displacement of the surface is observed through a longitudinal shift in laser spot position Δp = (p1 – p2). For practical matters, the errors with a triangulation-based laser scanner come mainly from the estimate of p, through δp. Error propagation gives the uncertainty in Z as: