- 125 - 
3.5. Summary 
The slightly lower correlation between the two profile sets in 
comparison with the profiles collected at Ashby Castle is 
considered a reflection on the areas where photogrammetric 
data capture was difficult due to shadow or lack of surface 
correlation. It is clear that the laser scanning data has been able 
to provide a very clear profile in areas where photogrammetric 
points were unable to be collected. 
The density of control points used for the laser scanning was 
much lower than that used to survey the hall by 
photogrammetry as the individual scans encompassed a wider 
field of view than the imagery and had no restriction on stereo 
coverage. 
Figure 7 shows a meshed model of the eastern face of the hall 
and an image taken from a similar orientation. Certain features 
are more apparent in the scan data due to the angle of 
illumination - for example the corbel highlighted in Figure 7 is 
clearly visible in the meshed scan data but not the image. The 
ability to interact with the data to perform such interpretation is 
a clear benefit of scanning. The meshed model does, however, 
still require some editing before it can be considered a true 
representation of the hall. 
Figure 7. A meshed scan model (top) produced by the Cyrax 
2500 and Cyclone software and a digital image (below) for 
comparison. Corbel is highlighted. 
4. DISCUSSION 
Imagery is currently the most common method for cultural- 
heritage recording. It provides not only a geometric record 
(with further photogrammetric processing) but also a qualitative 
record of the subject, ideal for interpretation and investigation. 
The products from photogrammetric surveys are accurate and 
efficient, and the archive value of imagery is well 
acknowledged; allowing measurement of a subject well after it 
has been damaged or destroyed. Imagery, however, is not a 
direct method of geometric data capture and in some instances 
may have shortcomings, such as in complicated areas of relief 
where detailed stereo-coverage may be required to produce 
measurements, or in areas of shadow where lack of image 
correspondence may require intensive manual interaction (the 
measurement/editing of data points) or a complete failure to 
measure any data. The secondary processing stage also 
lengthens the time required to produce geometric information. 
Laser scanning captures geometric data directly without the 
need for a secondary processing stage. The measurement of 
complex areas of surface detail is much easier with laser 
scanning than stereo photogrammetry - especially when 
dealing with sharp edges. Scanning captures a large amount of 
geometric data in a short length of time (typically 10-15 
minutes per scan), however, the data captured is simply a dense 
collection of points as opposed to discrete points of interest. 
No intelligence is present within the data without further 
processing. Points can be incorrectly measured due to multi- 
path or mixed-pixel effects and these points need to be 
identified to ensure they are not used for modelling or 
measurement. The resolution of a laser scan needs to be 
matched against the features of interest to ensure that those 
features are visible in the resulting point cloud - the scanner 
selected must provide a resolution greater than the smallest 
feature to be measured. This is perhaps the most important 
question for laser scanning within cultural heritage applications. 
As many recording projects are performed for archive purposes 
and the features of interest may be unknown at the time of 
capture what is an appropriate resolution to scan at? 
The desired resolution may, in part, be determined by the 
required product. Presently, this product may take the form of 
a meshed model and cross sections but will not be able to 
replace the advantages of image based products: high resolution 
archive data, ease of capture etc. It could, however, provide the 
surface models for orthorectified photography eliminating the 
need for lengthy photogrammetric surface extraction and 
editing. The interpretational value of a meshed 3D model with 
the ability to alter the light sources in real-time may also 
provide a new product opportunity. 
Although in both of the projects described here registration 
between scans was performed using targeted points, registration 
of point clouds can also be performed through matching 
techniques, such as the Iterative Closest Point algorithm (Besl 
and McKay, 1992). This approach minimises the need for 
control points, however, some care would be required in 
providing sufficient redundancy for quality assurance checks. 
The matching of points in this way is similar to the use of 
surface matching within photogrammetric applications for the 
orientation of stereo-models (Rosenholm and Torlegard, 1988). 
5. INTEGRATION OF SCANNING INTO THE SURVEY 
WORKFLOW 
5.1. Desired result 
The justification for the use of any new technique must be in its 
ability to provide added value and for better efficiency. 
Scanning is not, realistically, able to replace the current use of 
image based techniques at the present moment. However, it 
may improve the value and efficiency of survey work if 
scanning was used in conjunction with current techniques, in