Data produced by 3D laser scanners are typically large “point clouds” of X, Y, Z data of objects and sites that range from the very small (a tooth) to the very large (an entire community). In addition to XYZ data, scanners often return Intensity (I) or color (RGB) data as well. The use of laser scanners is growing in heritage applications and there is a growing body of literature on their technical and operational characteristics. In the data provided, laser scanning has been the primary method used but a wide range of other approaches have also been used. In the following we provide an overview of some of these methods. More specific details relating to each project and data set are provided with the data.
Time of Flight (TOF) Laser scanning
When considering larger areas (e.g. 100m across and larger), systems using time of flight (TOF) as the measurement method are used. Other scanning methods including phase comparison and triangulation are used for smaller objects with more precise measurement. These other techniques are limited in their range to typically less than 10 m. Unfortunately TOF systems have the necessary substantial range but do not support the very high accuracies provided by the phase comparison and triangulation systems. In a TOF system a laser pulse is sent out and a portion of the pulse is reflected from any surface encountered. The distance to the surface is calculated from the time of the flight of the pulse, hence the name. In most systems highly precise mirror systems control the horizontal and vertical angles of the beam causing the pulse to move across the surface in a regular manner. The angles of the mirrors are used to calculate the X and Y values for each pulse. Each system then produces a data set that consists of multiple XYZI values – where I represents the scanner intensity value. As the pulse travels greater distances from the instrument it expands. This presents some specific challenges. A related issues is that many measurements involve the pulse interacting with the surface at low, glancing (oblique) angles. This is particularly the case with large areas where the scanner cannot be located on a higher location relative to the area. Further, since the scanner can only measure surfaces that are illuminated by the instrument any surface that is obscured is in the measurement shadow. To remedy this it is often necessary to perform multiple scans from different vantage points and integrate the results. Such integration can be very challenging. Scanning of large areas creates very large point clouds. In our work numbers exceeding 200 million are not uncommon. Commercial software systems, even those specifically designed for processing these HDS data, are challenged to manage these numbers of observations. Our research group is at work leveraging new developments in 3D object-relations databases (esp Oracle 10G Release2) to address the challenges that large sites and large data sets present. A final challenge in the processing of these large data sets is the extraction of relevant information, such as CAD elements representing architectural features of interest. Work is underway in the areas of automatic and semi-automatic feature extraction as well as new developments that add spectral data (e.g. true color, CIR, thermal, etc) to the X, Y, Z information (e.g. XYZIRGB) and then use strategies derived from image segmentation strategies to automatically identify features.
Portion of laser scanning point cloud data with zoom area delineated
Point cloud data for “zoomed” area
Triangulation and phase comparison based laser scanning
(This section under construction)
Traditional Photogrammetry for point cloud generation
A final technical approach that can be used to develop HDS data is via traditional photogrammetric methods. There is a substantial body of literature on this approach and we will not review it in any detail at this time except to note that new methods are being developed that generate point cloud data from the photogrammetry. In the past it was commonly the case that any measurements required were extracted from the metric stereo-images using electro-mechanical analytical plotters. These plotters have been replaced by modern digital photogrammetry software. In general measurement extraction via photogrammetry requires qualified professional users but there is currently substantial activity in the creation of automatic measurement.
Some technical issues in long distance laser scanning
There are some specific characteristics of laser scanning systems when used in large area survey that influence the results and the most effective field approach. Scanner use is influenced by two central aspects of the geometry of the systems. The first is that the area capable of being measures is that which can be illumed by the laser. Any location in “laser shadow” has no data. Conversely an object in front of another casts a no-data shadow on it. The second is that, although all the systems send out a very constrained pulse of light, the inherent properties of light transmission means that the pulse expands in its diameter as it moves outward from the instrument.
The implications of the laser shadow effect mandates that almost all large areas and structures will require multiple scans to cover the entire area. The number and location of these scans will depend on the complexity of the surface and/or architecture. Take the case of a simple cube – at least four scans (and more likely more) would be needed to cover the sides and perhaps a fifth for the top. Introduce a second cube before the first and the scan requirements become much more challenging. Related to this issue is the density and precision of the scanning needed. Each scanner has an effective field of view – or the area that can be illuminated by the instrument. There are normally different values for the vertical and horizontal capabilities and both are quite variable ranging from a low of 40 degrees to 360 degrees. With an instrument with a 40 degree field of view, for example, it may be necessary to set up the instrument in multiple locations to cover an area. A wider field of view would require fewer individual scans. The situation is, unfortunately, not that simple, however. If the field of view is large and the distance to the surfaces of interest also large then surfaces at the edges of the field of view can be substantially farther away and at much greater oblique angles to the beam. Both situations can affect the quality of the results. In many situations, even with instruments with quite substantial fields of view, it may be appropriate to reduce the field actually used and do multiple scans to minimize error.
The challenge of laser shadow is a critical one for large area high density scanning and should be clearly considered in field planning. It is important to preview each scan to identify unexpected shadow areas in the field to insure that all areas have been covered. Where there is substantial relief or any structures present will almost always be necessary to have scans from opposite sides to insure that areas are not overlooked. If access to all sides of a monument is not feasible it may be possible to set up the instrument in an oblique location and acquire the data. Consider a flat wall that is across a small alley from another structure. It would be impossible to image the wall fully from one location. In this case either multiple small scans perpendicular to the surface could be made or fewer and larger oblique ones. In the case of the oblique scans, however, the geometry of the acquisition means that the resolution (and resolvable details) will vary more across the area scanned than it would with a perpendicular scanning case. Processing of oblique scans present a number of challenges and, in general, it is recommended to avoid oblique scanning wherever feasible.
Beam expansion is an inherent property of any light propagation. The impact becomes greater, however, as the distance to the surface illumed increases. The following illustration shows the different beam diameters for various distances for the Optech ILRIS scanner used I a number of the large area scans. The formula for the Optech beam width is:
Diameter in mm = 0.17 * Range to target in m + 12
The spot size at 500 m is 97 mm. Sizes at various distances are shown in the figure above.
The spot size effect is shown graphically in the figure below where spots of various sizes are superimposed on a surface with reference points spread across the surface at a nominal 20 mm.
Optech beam spot sizes overlain on 20 mm spaced data points.
The impact on the beam size and resolution is important but complex. In order for a distance measurement to be made only a portion of the pulse needs to be reflected from the surface to the instrument. As the distance becomes greater the strength of the pulse is dispersed and the returning energy level is similarly reduced. Highly reflective surfaces return more energy so it is possible to measure them at greater distances that less reflective ones. The beam's energy is in the infrared range (on most systems) and infrared energy reflects somewhat differently than normal light. The structure of the surface also impact the energy reflected. A diffuse surface may scatter the beam for example. For the Optech systems, for example, the instrument can operate at a range of 350 m from a surface that is only 4% reflective. At 8,000 m the surface needs to be approximately 20% reflective or better.
Each instrument will have its own properties but this general pattern will always apply.
In addition to the increases in beam size at distance size there can be complex beam interactions with various surface geometry. The below figure illustrates the ideal case. Where the instrument is perpendicular to a flat surface. The majority of the instruments provide a first and last return. As the example shows both the first and last return are the same and both are coincident with the surface.
When the beam strikes a surface at an (increasingly) oblique angle the process and resulting measurements become much more complex to interpret. The following illustrations show this complex process. In the first example the beam is striking a rounded corner at a moderate angle. The system will use the center of the beam to calculate the X and Y values. The first return from the surface, which will typically be used by many systems as the defining distance, will closely correspond to the actual distance. If the last return is used, which can sometimes be the case, the Z value would be substantially farther away than was correct.
Some return properties for rounded corner interaction.
In the third example the beam is striking a surface with a sharp edge at a somewhat oblique angle. The X and Y values of the wall are correctly shown by the center ray while the Z value will not be the wall surface but will be based on the first return from the corner. Causing the wall point to appear to be closer (if first returns are used) that in the case. If last returns are used (some instruments provide for this option) the wall point will have a correct X and Y value but a Z value that places it at a greater distance than is correct. If the beam were moved further down the wall the Z error would actually be increased as shown in the next figure.
The first and last return values shown in most of the drawings are actually only theoretical in that there would be insufficient energy returned from the far margins of he beam so the actual variation would be somewhere within these limits. Furthermore the surfaces in the examples are shown as uniformly reflective and smooth. This will not be the case in the real world, and particularly not true for heritage sites. It is possible to formally calculate these values for various geometric situations but it is important to note that the surface c
Beam angle encountering a sharp corner and the differing Z values that result.
Beam and return relationships for oblique location on a wall surface.
In many field situations these complex process can be avoided by a detailed consideration of the instrument location relative to the surfaces scanned.