|
 Scanning
Methodology
The
ATL is equipped with a Minolta Vivid 700 Non-Contact Laser
Digitizer, a Dell 610 dual-processor workstation, and Polyworks
Modeler software that allows the lab to produce digital 3D
surface models of physical objects. This equipment will be
referred to as "digitizing equipment." When the object is
an archaeological artifact, the lab refers to the 3D surface
model as a "digital artifact." What follows is an explanation
of the process (to which we will refer as "digitizing") whose
final product is a digital artifact. Once a digital artifact
has been produced from a physical archaeological artifact,
it is said to have been "digitized." Before an artifact is
chosen to be digitized, the setup and placement of the digitizing
equipment must be carefully considered.
The
Vivid 700 requires at least 60 centimeters of distance between
the lens of the digitizer and the object, and can accurately
digitize an object up to a meter from the lens. Any object
placed in front of the digitizing lens that is within 60 centimeters
will cause the scanner to default to safety mode, and not
scan. This is because the light emitted from the Vivid's laser
beam could cause damage to the eyes of a person at close range
if such a close proximity to the digitizer were permitted.
Thus a work area of 1.5 square meters is required for the
Vivid 700 to function properly.
The
Vivid 700 is connected to the 610 workstation via a SCSI II
connection. This allows for the fast transfer of digitized
data from the Vivid 700 to the Vivid data-capturing software
application on the workstation. The workstation is setup to
the rear of the Vivid 700 to avoid interfering with the laser
system of the digitizer and forcing it into safety mode.
Once
an artifact has been chosen to be digitized, it is placed
in front of the Vivid 700, approximately 60 to 80 cm out from
the center of the digitizing lens, and the Vivid data-capturing
application is started on the workstation. The artifact is
positioned on a supporting medium in order to hold steady
its orientation and distance in relation to the digitizing
lens. It is at this point that the human operator selects
"One Shot" from the "Remote" pull-down menu. The software
when then activate the digitizing lens causing a live video
feed to be displayed on the Vivid 700's rear monitor screen.
At the same moment a still image of the artifact is displayed
on the workstation's monitor in the "Remote-One Shot" pop-up
box. The operator then ensures that the live video feed of
the artifact's image is centered in the digitizing lens by
moving the artifact perpendicularly to the digitizer. At this
time, the operator also has the option of using the zoom feature
to enlarge the artifact within the rear monitor screen. When
the operator is satisfied with the orientation and positioning
of the artifact, he/she then presses the "release" button
in the "Remote-One Shot" pop-up box.
When
the release button is pushed, the scanner initiates the laser,
which emits a laser beam that is reflected off a mirror causing
the laser beam to spread out from a point reflection to a
horizontal line reflection. As the mirror rotates the laser
beam reflects back from the artifact, as a red line appears
to travel from the top to the bottom of the artifact. This
process can be referred to as "painting" the artifact. As
the reflected laser beam returns to the digitizing lens, the
Vivid 700 calculates the three-dimensional(3D) location of
each point at which the laser began its return trip to the
digitizing lens. Once the laser beam line reaches the bottom
of its reflecting range, the laser and mirror are returned
to their original states, ready to paint the artifact again
if necessary. Lastly, the digitizing lens takes a digital
photograph of the artifact. The 3D point data and its associated
digital photograph are then compiled by the Vivid 700, and
sent to the Vivid data-capturing application running on the
workstation. This compiled 3D data and digital image is saved
as a *.cam file(camera data file) at this time, is referred
to by the ATL as a "scan."
Once
in the Vivid data-capturing software, the scan is turned into
a 3D polygonal mesh. This mesh models the contour of the area
of the artifact that reflected the laser beam line back to
the digitizing lens. The polygonal mesh also has associated
mapping coordinates so that the digital photograph generated
by the digitizing lens can be viewed as if it were reflected
as light from the actual physical object. The data-capturing
software has the ability to perform several editing functions,
although most of the data editing is performed in the Polyworks
Modeler package. After a few simple edits to remove obviously
bad data, the data-capturing software saves the 3D mesh as
an element in a *.vvd file format. This file can then be imported
into a Polyworks Modeler module named IMAlign.
After
one area of the artifact has been digitized, the artifact
is rotated so that a second area can be digitized. The areas
of digitization should overlap slightly so that corresponding
points found on each scan can be oriented to align to the
original contour of the physical artifact. The process is
continued over and over until the scan digitally recreated
the entire surface contour of the physical artifact. Each
*.vvd file is thus created after its associated scan has been
brought into the data-capturing software and simple edits
performed its associated 3D mesh. Then after each *.vvd file
is created, it is imported into the IMAlign program to be
aligned with each of the other *.vvd files representing the
surface of the physical artifact.
Once
the *.vvd files have been aligned to recreate the entire surface
contour of the physical artifact, the files are merged using
a command-line module of Polyworks Modeler named IMMerge.
This command-line module then outputs a model of the merged
*.vvd files in wide variety of 3D model formats such as *.pol
(Polyworks format), *.wrl(VRML 2.0/97 format), *.obj(Wavefront
format), *.stl(STL milling format), and *.igs(IGES a CAD/CAM
format). The ATL prefers to output merged models in *.pol
format for ease of use with Polyworks Modeler. Each model
can be saved in any of these 3D formats at any time during
the Polyworks modeling process, and the ATL outputs the file
version of each model in *.wrl format.
A
merged model is not a complete model, however, and must be
brought into yet another Polyworks Modeler module named IMEdit.
This model allows the operator to find and fill holes in the
digitized model using mathematic algorithms in an attempt
to match the filling contour with the area around it. IMEdit
also allows the operator to delete extra polygons, smooth
out the digitizer error (which is 300 microns for the Vivid
700), and perform both simple and complex measurements. The
measurement capabilities of IMEdit include surface area, circumference,
angle, cord distance, and volume. Once a model is properly
edited in IMEdit, it is a completed model, however the model
may be run through two more Polyworks Modeler modules, IMTexture
and IMCompress.
IMCompress
is a second command-line module that evaluates the shape of
a model for data that can be removed without changing the
overall shape and size of the model in 3D space. IMCompress
does this through the use of mathematical algorithms to compare
changes in the shape of the model's surface. In order to use
IMCompress, the operator must specify a tolerance that can
be as small as 1/1000 of a micron.
IMTexture
is a third command-line module of Polyworks Modeler. This
module is used to produced a bitmap texture graphic from a
completed digitized model. This bitmap can then be applied
to highly compressed models in order to show lost surface
features for educational purposes.
The
amount of time it takes to run an artifact through a complete
digitizing process varies from artifact to artifact and from
person to person. A well-trained operator working on one workstation
can complete an average digitized artifact in about four hours.
Two well-trained operators utilizing two networked workstations
(one operator producing scans, the other aligning and editing
models) would cut the time to produce a digitized artifact
by approximately half, and possibly more depending on the
detail of the artifact. A complex object, such as a highly
detailed endocast, may take as much as six hours to complete
for one well-trained operator.
These
estimates are based upon modeling work that the ATL has undergone
in the past year, and assume that the operators are working
on fast dual-processor workstations, running top of the line
editing software such as Polyworks Modeler. Before the use
of Polyworks Modeler, the ATL used the Vivid data-capturing
software to align and edit its digitized models. During this
nine month period, the ATL was only able to complete one satisfactorily
finished digitized artifact, whose process took approximately
32 hours. This same artifact was re-digitized using Polyworks
Modeler in approximately six hours by novice operator. The
re-digitized artifact was found to be a much more accurate
digitized model than the first.
| 