Figure 3. Profiles: Ashby Castle. 
123 - 
Figure 2. A sub-section of the intensity image produced by the 
Riegl LMS Z210 scanner. 
2.3. Photogrammetry 
As a comparison for the laser scanned data, photogrammetric 
data capture was also performed at Ashby Castle using a film 
based Wild P32 metric camera. Photography was taken at an 
average photo scale of 1:350 and was scanned onto Kodak’s 
Photo CD format to allow it to be used in LH Systems SOCET 
SET DPW. 
2.4. Profiles 
The laser scan data was used to produce two longitudinal 
profiles through the façade. Profile 1 was placed across the left 
wall-end-section and comprised of an area of rough stone work. 
Profile two was located on the main façade across a small 
recessed window. Profiles were also collected using the 
scanned P32 imagery in SOCET SET. The recessed window 
area was in shadow at the time of photography and the second 
photogrammetric profile had to be broken at this point. Figure 
3 shows the location of the two profiles. Figure 4 shows the 
two profiles using both scan (points) and photogrammetric (line) 
techniques. To provide an indication of the similarity of the 
profiles the laser scanner data was sampled at the collected 
photogrammetric data points, followed by the calculation of 
correlation coefficients. The profiles showed a high level of 
agreement with correlation coefficients of 0.99 and 0.97 for 
profiles 1 and 2 respectively. 
Figure 4. Profiles one (left) and two (right) from Hastings 
Tower. 
2.5. Summary 
It is clear that the laser scan data has been able to provide data 
in the recessed window area in profile 2 where 
photogrammetric data capture was not possible due to shadow. 
However, although the point cloud contained a large number of 
points, the resolution of the scanning was found to be 
insufficient for the interpretation of small features of detail. 
This is especially the case for features at long distances from 
the scanner such as the fireplace at the top of the tower. It was 
also noticed that some data points were clearly incorrectly 
located, most notably at edges in the scene possibly due to 
multi-path or mixed-pixels affects. These points would need to 
be identified (preferably automatically) and removed from the 
data cloud in order to ensure they were not used in the final 
product.