CIPA 2003 XIX th International Symposium, 30 September - 04 October, 2003, Antalya, Turkey 
622 
responsible for controlling the narrative accuracy of TLS, 
while point accuracy has the greatest impact on geometric 
accuracy (although intuitively both point density and point 
accuracy will effect each type of accuracy). 
In previous photogrammetric/topographic specifications the 
required accuracy of a measurement has been determined by 
scaling the width of the standard output line width to 
calculate the required “real-world” accuracy. For example 
for survey output at a scale of 1:20 and a standard line width 
of 0.18 mm, an accuracy of 3.6 mm would be required. 
Originally this was applied when hard copy output was 
required. However, with most deliverables from survey 
projects likely to be digital this assumption is less 
appropriate as data could be plotted at potentially any scale. 
The use of scale however still provides some control as to the 
use of the data, providing the user with information relating 
to the accuracy of the information. Scale was also useful in 
suggesting the practical dimensions of objects for any 
particular scale - based on the size of a standard A1 sheet. 
Figure 2 A background scan, top, and a detail scan, 
below, of the area marked with a rectangle. 
The previous metric survey specification however defines a 
different level of accuracy when using different techniques, 
despite requesting survey at the same scales. In order to 
maintain continuity with previous measurement practises it 
was decided to adopt the same required level of geometric 
accuracy as that required for building survey. By applying 
this level of precision to scan data the use of the majority of 
laser scanners operating over 2 m is unsuited to survey at 
1:10 scale, however survey at 1:20, 1:50 and 1:100 scale are 
all possible. The choice of an appropriate point density is a 
subjective decision and although it is acknowledged that as 
high a point density as possible may be desirable this must be 
balanced with storage requirements. 
It was also important that the specification acknowledged the 
practical constraints on the data that can be captured. For 
example it is difficult to scan every portion of a complex 
structure, especially when scanning from ground level where 
it is not possible to record elevated window ledges or 
doorways in full. Where such areas are required it should be 
noted in the project brief and it may be necessary to amend 
the data collection process or investigate other methodologies 
all together. 
Scanning in heavy rain is deemed unsuitable, mainly due to 
concerns of having water droplets on the scanning window 
which may cause diffraction of the measurement beam, 
resulting in erroneous measurements. 
The likely need for two levels of recording is also 
emphasised. Background scans are appropriate for the 
spatial location and orientation of areas of detail which are 
scanned at a higher resolution, Figure 2 gives an example of 
using detail and background scanning using a Cyrax 2500. 
The process of registration differs between each TLS on the 
market. Some systems opt for the conventional survey 
approach of using a known reference object and known 
instrument position while other systems use a resection 
solution as performed in photogrammetry. It is necessary to 
specify for both methods of registration as each has a valid 
pedigree. For resection calculations the residuals of the 
estimated parameters are required to meet the accuracy of 
individual scan point measurements. Where registration is 
performed using a known scanner location and a known RO 
the precision of the scanner coordinates must be shown to 
equal the standard requirement of 4 mm for an XYZ control 
point as defined in the current metric survey specification for 
English Heritage. Cloud matching routines are deemed 
suitable, providing additional targets are used to confirm the 
accuracy of the procedure meets the specification. 
A particular feature of TLS, especially for cultural heritage 
survey, is the amount of additional information captured 
during the survey process. This includes imagery, sketches 
and notes. Copies of any data captured on site to aid the 
survey process are required on delivery to ensure any 
subsequent processing has all the available information. 
Where specific additional information is required it will be 
noted in the project brief. Imagery showing the scanned area 
is required for all surveys and to ensure any imagery is useful 
for interpretation this must have a resolution of at least 1500 
x 1000 pixels. 
As most sites in the UK under the care of English Heritage 
are open to the public, the safety of visitors is vital. Lasers 
have a justifiable reputation for being dangerous. A set 
standard is required to define the necessary precautions that 
should be taken. The specification adopts the classification 
and requirements of IEC 60825 (IEC, 2001). All laser 
classifications should be quoted to this standard. It is noted 
that at least three different specifications for laser 
classification exist; the European IEC standard; BS 
60825:1994 and the American ANSI standard (ANSI, 2000). 
Classification between these specifications is not consistent