The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing 2008
writers of this paper measured approximately 500 tooth replicas
using a mechanical stylus method, (Mitchell et al., 2003). Such
a procedure is clearly very protracted and inefficient. Many
other dental surface researchers carry out similar measurements,
(see, e.g., Azzopardi et al., 2000; Bartlett et al., 1997; Mehl et
al., 1997; Pesun et al., 2000; Pintado et al., 1997 ).
1.4 Goal
It is seen as highly advantageous for dental research to be based
on direct intra-oral measurement. Firstly, direct intra-oral
measurement would overcome the tedium of the casting
procedure for patients as well as dental workers. Secondly, the
use of photogrammetry could improve the efficiency of the
measurement 'procedure, as has been done for replica
measurement (Grenness et al., 2008).
1.5 Alternatives
Various techniques for dental replica measurement are
imaginable. However, for intra-oral measurement, any
instrumentation must meet severe access criteria, especially if
any side of the tooth - not simply the most easily accessed - is
to be measured. Electron microscope methods fail in this
regard, as indeed do other methods involving microscopes.
Among the optical methods, the obvious alternative is laser
scanning but it faces a far high equipment expense than does
photogrammetry. Other optical options include shape-from-
shading, moiré fringing and structured light techniques. The
feasibility and value of optical intra-oral measurement is
signified by the CEREC (Sirona GmbH, 2008) and D4D (D4D
Technologies, 2008) intra-oral measurement systems, even
though they measure using optical but not photogrammetric
principles. The cost of optical dental measurement techniques
based on structured light patterns is hard to estimate, because
the commercial units are sophisticated and moreover are
available only when linked to crown machining software and
hardware.
1.6 Photogrammetry’s advantages
Intra-oral measurement by optical methods seems to be a
feasible option. Photogrammetric measurement can be
executed at low cost, because the only significant hardware
requirement is a camera. Intra-oral cameras are commercially
available. Although there is a likely requirement for ancillary
items such as lighting, lenses and mirrors, these are relatively
cheap. The other significant photogrammetric components are
software. Given the need for automation, and given the current
sophistication of automated image matching, photogrammetry
was seen by the writers to be deserving of some
experimentation.
1.7 Requirements
The above description suggests that there is a demand for
development of a tool which could be used to measure any
surface of any tooth in the mouth to aid any dental research
which involves studies of the loss of the external enamel
surface of the tooth. However, to replace alternative and
existing procedures which involve taking castings, and perhaps
replicas, of the tooth for subsequent measurement by
mechanical or photogrammetric means, as is intended for such
purposes as studying tooth surface loss caused by decay and/or
erosion, the technique must be superior in other aspects as well.
The requirements are therefore seen to be:
Able to measure any surface of the tooth, this defining
access requirements.
Quick, comfortable and safe for the patient, i.e. easier
than taking castings which can take a few minutes.
Simple and easy for the dentist and/or researcher, and
accordingly the photogrammetric measurement must be
entirely automatic.
Providing an absolute accuracy of about 0.01 mm in
absolute terms. This accuracy represents a low relative
accuracy of 1:1000 if the tooth is about 10 mm across.
2. THE PHOTOGRAMMETRIC DESIGN
2.1 Imaging
The problem of access to all teeth can be overcome by using
commercial intra-oral dental cameras. This work is based on a
Flexiscope Piccolo camera, providing colour imagery at 768 x
576 pixels size. Collecting suitable multiple images is a serious
difficulty. If a single camera is used, movement of the camera
in close proximity to the human patient while controlling
translations and rotations of the camera relative to the patient
can be difficult. The use of two cameras is therefore attractive;
Grenness et al. (2008) decided that “close range applications,
by their compact nature, are well suited to fixed base stereo
camera equipment. If the stereo camera is sufficiently rigid then,
once fully calibrated, object space control is not required".
However, the use of two cameras can severely limit access to
certain regions in the mouth, and, moreover, the use of two
separate cameras complicates image file handling, and this has
also been seen as undesirable for this project. Good estimates
of parallax values are also needed to get reliability and
matching success, and this demands careful control of the
camera position.
2.2 Calibration
The Flexiscope camera lens has a short but unknown focal
length with a 105° of field of view (Inline, 2008) and, not
surprisingly, high distortion levels, and it demanded calibration.
Figure 1. Intra-oral camera image of calibration object, only
approximately 10 mm in size. Distortion is apparent.
Calibration is problematic with small objects. A determination
of both principal distance and distortion parameters has been
undertaken using a small in-house test object, approximately 10