During a series of preliminary investigations, optical texture was successfully incorporated into epoxy resin tooth replicas. A semi-metric film camera fitted with a 100 mm macro prime lens on bellows and a further 50 mm extension was shown to provide suitable images for future work.
Chapter 5
Stereo-photogrammetric mapping of tooth replicas
incorporating texture
A manuscript prepared for submission to the Photogrammetric Record
Published in Photogrammetric Record, 20(110): 147–161 (June 2005) M. J. Grenness, The University of Tasmania,
J.E. Osborn, The University of Tasmania, M. J. Tyas, The University of Melbourne, Recipient of E. H. Thompson Award,
(Remote Sensing and Photogrammetry Society, UK, September 2007)
Abstract
Commercial digital photogrammetric software has been applied to
convergent stereoscopic photography of human tooth replicas prepared to exhibit optical texture resulting in successful generation of 3D coordinate data. Tooth replicas were imaged using a semi-metric 35 mm camera and f = 100 mm macro lens on extension bellows. Model precision was within
acceptable limits of 12 µm or better for manual target matching and 21 µm or better for automatic image matching. Further improvement in optical texture is required to achieve automatic image matching precision comparable to that of manual target matching. Small errors in interior orientation
parameters attributed to instability in the bellows as well as small errors in the relative orientation resulted in some systematic errors. The use of a fixed camera lens system is expected to reduce these errors. When combined with commercially available, moderately priced, digital SLR cameras this brings 3D model generation closer to everyday clinical dental practice.
5.1 Introduction
In dental research and clinical practice 2D representations or images are frequently required in order to aid diagnosis, analysis, communication with third parties, and for documentation and educational purposes (Haak and Schirra, 2000; Phelan, 2002). The acquisition of 3D models and data has, with some notable exceptions, been largely confined to the research arena.
Generation of 2D images and 3D models for dentistry fall into three broad categories: facial, full dental arch and single tooth.
Imaging of single teeth is relevant to the documentation, monitoring and study of tooth surface loss (Xhonga et al., 1972), restoration performance and degradation (Roulet et al., 1983), 3D finite element analysis modelling
(Palamara et al., 2000) and for the mapping of prepared teeth for computer- aided-design/computer assisted manufacture (CAD/CAM) restorative processes (Leinfelder et al., 1989). Laser scanning of tooth replicas has
become commonplace since the mid-1990s (Azzopardi et al., 2001; Folwaczny et al., 2000; Mehl et al., 1997; Perry et al., 2000). Traditional photography has been used to aid in the diagnosis of tooth surface loss (erosion) on anterior teeth with limited success (Al-Malik et al., 2001), photography of light-
induced fluorescence to detect dental caries has been trialled (Buchalla et al., 2002), and cross-polarised photography has been used to aid in the detection of enamel defects (Robertson and Toumba, 1999). Laser scanning of prepared teeth as a part of CAD/CAM procedures in the dental laboratory has become commonplace (van der Zel et al., 2001). A system (CEREC; Sirona Dental Systems GmbH, Bensheim, Germany) based on structured light and a charge-coupled device (CCD) camera has been developed for the capture and generation of 3D coordinate data from prepared teeth directly in the mouth as a part of a CAD/CAM system for the production of milled
restorations. Accuracy of 25 µm is reported by the manufacturer (Leinfelder
et al., 1989; Markowski, 1990; Mou et al., 2002; Parsell et al., 2000), although this may be reduced significantly due to the requirement for powdering the
tooth prior to imaging (Kurbad, 2000). Stereo- photogrammetric applications have been reviewed by (Chadwick, 1992) (for example, (Chadwick et al., 1991; Clarke et al., 1974; Lamb et al., 1987; Mitchell et al., 1989); however , the method has not found general utility or acceptance. Accuracies for all
methods of 4 to 20 µm are reported (Chadwick et al., 2002; Mehl et al., 1997;
Peters et al., 1999).
There is a need for high accuracy, in the order of 10 to 50 µm, in clinical
mapping of teeth in order to monitor the development and progression of tooth wear and surface defects and lesions. This would enable better clinical decisions about the need and effectiveness of potential therapies, varying from the use of remineralisation preparations, surface coatings and varnishes through to restoration of the tooth surface with long-lasting restorative materials. There are presently no standards or criteria for accuracy and precision of clinical tooth mapping for monitoring purposes. A recent review of the techniques to measure tooth wear and erosion (Azzopardi et al., 2000) concluded that all current techniques were confined to the laboratory and that there was a need for a simple, reliable technique for clinical use. Digital photogrammetric techniques can produce accurate 3D models using
traditional photography provided adequate surface optical texture is present on the surface of the object (Mitchell et al., 1999). Optical texture can be employed using structured light and a single camera (Leinfelder et al., 1989), unstructured light and two cameras (Dirksen et al., 2001), or targets applied in a thin film to the surface (Osborn and Wise, 1996). The incorporation of targets into the material used to produce tooth replicas has not been reported.
This paper describes the incorporation of optical texture incorporated into single tooth replicas, the use of digital stereo-photogrammetry to generate 3D surface models of the replicas, and presents the results of accuracy and precision testing.