|
|
3D Imaging
'Real World' 3D Capture
3D Capture Basics | 3D Capture Methods | Evidentiary Chain of Custody
3D capture is the acquisition of 'real world' 3D information that is intended for use in a digital environment. Information describing shape, color, and surface reflectance properties can be recorded. These recordings may then be reviewed digitally or imported into a virtual world.
3D Capture Basics
3D capture acquires empirical documentation of objects and sites. Two major types of acquired data are 'point clouds' and color information.
| |
Point clouds are sets of point locations composed of cartesian ordered (X,Y,Z) triplets. For example, (0,0,0) is the location of the coordinate axes Origin. The information contained in point clouds is called 'range' information. Point clouds are the most basic form of 3D geometry. Once collected, the point cloud can be displayed as a solid shape or as a dense mass of individual points. The range information contained in the point cloud cam also be used to build more complex geometric structures such as polygon meshes and parametric surfaces.
|
| |
Color information, associated with each (X,Y,Z) point, can be acquired by several 3D capture methods. This color information comes in the form of numerical values recording the amount of red, green and blue, the three elements of projected color. The three color values are organized into an (R,G,B) ordered triplet.
A useful measure of point cloud resolution is 'sample density'. Sample density counts the the number of points per unit of area or distance, as in one sample every centimeter or 64 samples per square millimeter. The higher the sample density, the thicker the point cloud. |
3D Capture Methods
Laser Scanning | Structured Light | Photogrammetry | Contact Digitizers
There are a number of ways to capture 3D data. Several of the most commonly used methods are examined below:
| |
Laser Scanning |
| |
|
Laser scanners record point locations by projecting a low intensity laser beam onto the scan subject and then recording its reflection back to the scanner. Hundreds of thousands to millions of point samples can be acquired from each scan.
There are four examples of laser scanning systems in this site's Web Resources section. Three use monochrome lasers, Cyberware, Konica, Minolta, and Leica Geosystems, while one, Arius3D, uses combined red, green, and blue lasers.
|
| |
Structured Light |
| |
|
Structured light scanning projects organized patterns of white light, for example grids, dots, or stripes, on a scan subject as it is simultaneously photographed using a digital camera that is specially aligned with the projector. The photographic image shows the distortion of the projected light pattern as it strikes the scan subject. Associated structured light software processes the projected light pattern distortion disclosed by the digital photographs to generate point cloud data. Color information is inherently registered with the point cloud's range information because the range information is created from the color information.
There are two examples of structured light systems from Eyetronics and GOM mbH, and two examples of their use during the imaging of cultural heritage material in this site's Web Resources section.
|
| |
Photogrammetry |
| |
|
Photogrammetry uses optical characteristics to calculate the the 3D locations of the photograph's pixels. This method uses two or more photographs of the same subject, taken from different viewpoints. Several point locations on the scan subject that can be identified in two or more photographs are indicated on each image. These shared indicated points, coupled with the optical attributes of the images, enable the extraction of a point cloud's range information in software. As with structured light scanning, the photographically derived color information is inherently registered with the point cloud's range information because the range information is created from the color information.
|
| |
Contact Digitizers |
| |
|
Contact digitizers use a probe to physically touch the subject and record the location at the point of contact. Many contact digitizers position their probes at the end of articulated arms. These arms usually have joints with multi-axial rotation, enabling the probe to contact the subject's difficult to reach places. The multi-axial devices that support and track the probe are called 'Coordinate Measuring Machines' (CMM's). CMM's can reach sizes sufficient to scan airplanes. CMM's are often used to support laser, structured light, and photogrammetry scan heads and cameras. The CMM captures the position of the scanner or camera as the scanner or camera captures the scan subject. |
| |
Evidentiary Chain of Custody |
| |
Merging Multiple Point Clouds | Registering Geometry With It’s Texture |
| |
|
Cultural heritage documentation, and scientific visualization require a clear empirical account that assures users that the features on the virtual object accurately represent the features of the scan subject. A person examining a virtual paleolithic stone tool wants to have confidence that its features are on the original. |
| |
|
|
Merging Multiple Point Clouds
Given that most scan subjects require multiple scans from different positions to achieve full surface coverage, the ability to give a clear methodological account of the process of merging multiple scans is of great benefit for cultural heritage documentation, and scientific visualization.
Many point cloud processing software packages use a method of 'best fit' calculation to merge multiple scans. Best fit software compares the the point clouds' shape characteristics and tries to align them together with a minimum amount of separation in overlapping areas. While often quite accurate, this accuracy is variable and decreases in overlapping areas with little or no change in surface features. The ambiguity introduced by this variability muddies the the ability to give a clear methodological account of how the results of merging multiple scans were obtained.
Another approach is to use a photogrammetry like process. This process uses the knowledge of scanner or camera positions and orientations, to determine the accurate positions of multiple point clouds captured from different locations around the scan subject. This method offers a clear account of how the point clouds were merged. |
| |
|
|
Registering Geometry With its Texture
If (X,Y,Z) range and (R,G,B) color values are recorded together, the color information is registered exactly with its corresponding 3D point. The scientific certainty of this correspondence is highly desirable in cultural heritage work. 3D capture methods that generate point clouds from color digital photographs and lasers using red, green, and blue wavelengths together can give the clearest account of registering geometry with its texture.
Methods that map a calibrated video or digital image to the geometry are as good as the accuracy and reliability of the camera's calibration with the point cloud scanner.
When color information is acquired independently form the range information, as from a separately taken digital photograph, the color information must be mapped back onto the 3D geometry. The process of mapping the independently acquired color data back onto 3D geometry is an inexact science because of ambiguities introduced when determining where to map the color. Methods that project uncalibrated texture images onto scanned geometry use an iterative method of projecting the texture, examining the results, adjusting the projection's aim, and projecting the texture again. This trail and error method attempts to position the image properly on the geometry. While it has been demonstrated that intensive manual fiddling with texture map projections can yield fine results, 'eyeballing' offers a poor methodological account of registering geometry with its texture. This ambiguity introduced by the color mapping process is undesirable in cultural heritage documentation because it creates uncertainty about the truthfulness of the digital representation. |
| |
|
Ultimately, the requirement to provide an account of the evidentiary chain of custody will drive cultural heritage and scientific visualization 3D capture methods in the direction of those processes able to strictly account for their results.
|
|
|
|
