-10-
complicated signal analysis, the results may be more accurate
(at the expense of the measuring rate). Since a well defined
returning signal is needed, scanners using the phase comparison
method may also have a reduced range and tend to produce
more wrong or dropped points.
Laser/Receiver/
Time Measurement Unit
rates and real time processing are provided, however, these
instruments are an alternative to the scanning devices specified
above and may under practical requirements be considered as
3D scanners as well.
Object
Object
0
Lens
Figure 1. Time of flight principle
Figure 3. Triangulation principle: double camera solution
3.2 Triangulation scanners
Single camera solution. This type of scanner consists of a
transmitting device, sending a laser beam at a defined,
incrementally changed angle from one end of a mechanical base
onto the object, and a CCD camera at the other end of this base
which detects the laser spot (or line) on the object. The 3D
position of the reflecting surface element can be derived from
the resulting triangle (cf. fig. 2). This principle, too, has pre
decessors in surveying where range finders with constant bases
have been used. From there, it is also well known that the
accuracy of the distance between instrument and object
decreases with the square of this distance. Obviously, for
practical reasons, the base length cannot be increased at will.
Nevertheless, these instruments play an important role for short
distances and small objects where they are much more accurate
than ranging scanners (cf. fig. 4).
Figure 2. Triangulation principle: single camera solution
Double camera solution. A variation of the triangulation
principle is the use of two CCD cameras, each one at another
end of the base. The spot or pattern which is to be detected is
generated by a separate light projector which does not have any
measuring function (cf. fig. 3). A large variety of solutions can
be found. The projection may consist of a moving light spot or
line, of moving stripe patterns, or of a static arbitrary pattern.
The geometric solution is the same as with the one camera
principle, thus resulting in the same accuracy characteristics.
Not all devices using two cameras offer high point rates and not
all of them produce 3D coordinates in real time. If high point
4. ACCURACY CONSIDERATIONS
Accuracy is not always the predominant demand in cultural
heritage documentation. A standard deviation of a few
millimeters for a single scanned surface point is not of much
evidence if this point is part of an element possessing a regular
geometry (plane, cylinder or the like) and is just used to find the
parameters describing this element in a CAD representation. If
irregular surfaces have to be modeled (usually by a mesh
representation), noisy point clouds can be a rather nuisance in
processing, especially when the presence of edges does not
allow overall smoothing operations. Therefore, the scanning
procedure should be carried out with the most accurate scanner
available for the size and range of the particular object.
Since objects of many different sizes occur in cultural heritage
documentation, no single scanner can really be recommended
for all tasks. Instead, a selection of three different scanners co
vering roughly the ranges 0.1 to 1 m, 1 to 10 m, and 10 to 100
m is desirable. A large selection of scanners for ranges below 1
m is available (tab.l); a single point accuracy of 0.1 mm and
below can be expected. For the mid range from about 1 to 10 m,
the MENSI S10 and S25 (former names: SOISIC SD and LD)
triangulation scanners are presently the optimal choice (0.5 mm
at 2 m, 2 mm at 10 m, for the S25). For larger distances, again a
good choice of instruments can be found (table 1), yielding an
accuracy of a few millimeters to some centimeters depending to
some extent on their possible maximum range.
D istance
Figure 4. Scanner accuracy (Small parabola: triangulation
scanner with short base. Large parabola: triangulation scanner
with long base. Straight line: Ranging scanner)
