Laser Scanning and Photogrammetry: 21 st Century Metrology 
363 
4. ACCURACY AND INTEGRITY 
Baltsavias and Hahn (2000) promote the need for reliability indicators in the integration of spatial information. Redundancy is an 
important part of this process, it has an essential role in photogrammetry and surveying providing assessment of precision, ensuring 
the integrity of measurement. 
When scanning a surface, laser scanners provide a huge number of points that provide redundancy in the measurement of a surface, 
but redundancy is not present in the measurement of any one of these individual points. Each point has only a precision value 
propagated from the standard errors of the range and angular measurement. Riegl offer the ability to average scans providing some 
redundancy of measurement. Here each data point is considered to belong to one pixel on the scan image (see Figure 4 as an 
example). Based on the values of the horizontal and vertical angles it can be calculated which pixel in the image a particular 
measurement belongs to (Riegl 2001). By taking more than one scan of the object an average of readings for a particular 
measurement can be derived. 
Baltsavias and Hahn (2000) also state the need for tests on comprehensive datasets in cooperation with industry. The ability of 
scanners to meet survey type specifications must be investigated to allow appropriate guidelines to be produced on their use. Testing 
of all types of scanners should follow standardized procedures that produce comparable statistics. The selection of these tests should 
take into consideration the issues of accuracy, precision, resolution, response to different materials and software processing. Such 
testing may be as simple as the measurement of an object of known size at different ranges. A simple figure could be derived for the 
level of conformity the scanner shows to the shape of the object. 
Other approaches to the assessment of new techniques have used standardized datasets of particular monuments. One example is the 
CIPA Zurich City Hall dataset (Streilein et al., 2000), which provides test data for the evaluation of photogrammetric software. A 
similar archive would be beneficial for laser scanners, however this dataset should include scans of the same subject, for example a 
building, by different scanning systems. A user could then ensure particular features are evident before commissioning a particular 
scanner system on a project. 
5. DATA PROCESSING AND PRESENTATION 
Scanners are not restricted to recording XYZ data, many LiDAR scanners also record the intensity of the returned pulse. The display 
of point data with the attached intensity can improve interpretation of what may otherwise be a confusing scene. Both the Cyrax 
2400/2500 and the Riegl LMS-Z210 record intensity, and as this is based on the sensing of active energy, intensity data is also 
collected in areas of shadow, or in the dark. The LMS-Z210 also records full colour data based on the passive detection of light at 
the time of scanning which is useful for the display and presentation of data. Figure 4 shows a single scan made using a Riegl LMS- 
Z210 scanner, illustrating the intensity, range and full colour data. Note how this data is displayed as an image rather than the 
commonly seen “point cloud”, each pixel on the images has an associated XYZ position. Other scanners such as the Arius3D 
actively record colour using three laser wavelengths and so record “true colour”, regardless of the ambient illumination. 
This leads to a discussion on the processing and presentation of scan data. This applies to data from all types of scanners, either 
triangulation or LiDAR. The first procedure in processing scan data is registration. If the scanner has been moved between scans to 
cover an object in full, the scans need to be brought onto a common coordinate system. This can be accomplished via the matching 
of common points within scans and transforming to a base system allowing registered scans to be viewed as one dataset. This 
process requires scans to contain common points and in some cases it may be easier to register all scans to a common local site grid 
established by normal survey methods. This way scans that do not overlap can be used in conjunction with one and other. The use 
of discrete points requires some method of point identification. A common solution is the use of three dimensional shapes that can 
be reduced to a single point. The Cyrax system utilizes spheres, which once scanned (from any direction), can be reduced to a single 
point, the Soisic scanner from Mensi also utilizes spheres to register scans. Another solution is the use of flat reflective targets. The 
Cyrax system supports a process of scanning targets at a high resolution and centroiding to determine a single point. The use of 
reflective targets is also possible with Riegl scanners. 
More common in the registration of close range scans is the matching of surfaces, although this technique is also used in large object 
scanners. This allows overlapping areas of scan data to be matched together; the redundancy involved in this method may lead to a 
better solution than the use of discrete points alone. Surface matching has been implemented in some applications in 
photogrammetry (Rosenholm and Torlegard, 1988). 
The methods employed in turning a point cloud into a useful product vary and will depend upon the subject and application. Fitting 
CAD primitives to point clouds is a popular approach in some modeling and industrial applications. Software such as Cyra’s 
Cyclone and Mensi’s 3Dipsos software allows this. Beraldin et al. (1997) reports that algorithms used to rigorously fit planar 
surfaces to some areas of point data deliver a more precise determination than predicted. Many architectural and heritage subjects 
however are irregular and would not suit this method of processing, therefore it is more likely a mesh will be used to display the 
scanner data. Meshing is also an accepted method used to process data of small objects such as archeological artifacts. 
The meshing of scan data can include some control of the mesh density based on the complexity of the surface, for example large flat 
areas require fewer triangles than more detailed areas. This allows some control over the size of the model produced, retaining detail 
where required but removing unnecessary complexity in other areas. The planning of scanning projects should ensure sufficient data 
is collected to fully record a subject. However the use of a mesh to surface the scan data may result in small holes where data has not 
been captured, for example where reflectance is low or where practical issues prohibit satisfactory coverage. Further processing may 
therefore be required depending on the desired product. Any further processing will be based upon an operator’s judgment rather 
than actual data; having implications in the future use of such data, such as in building analysis. It is important that a detailed record 
of the processing carried out is retained within the project metadata.