UofK SCANNER DEVELOPMENT FOR REMOTE SITES

We list the phases of development of our remote technology in reverse order. Note that we have two lines of technology, one with a laptop to tether the scanner and one with no laptop. All are battery powered.

PHASE 4B: (January 2010) B&W Shoeprint Impressions in Mud, Underwater and in Snow.

It snowed in Lexington January 8, 2010. It was lightly snowing during the scan process. None of the technology was ruggedized but since it was done in my back yard, I just covered the laptop keyboard with a towel and took my chances.

Figure: Scanner setup and 3 shoe print scan. (a) Scanner setup, (b) scanning with green light, (c) scan with color texture, and (d) quality map.

Figure: Three shoe print scan. (a) Metalized surface and (b) depth encoded. From left to right is a right boot print (highly worn) and right and left shoe print (new shoes).

In 2006 we used a different scanner to capture a mud footprint that we created in our lab. We also scanned the print under 1 inch of water. We then forgot about the mud print for 4 years and rescanned the dried and cracked print. The boot print variations in 2006 while the mud was wet was about 3 mm and the dried one is now about 1 mm variations. This deterioration is probably due to drying but could also be caused by vibration and moving the print several times over the 4 years…we are not really sure?

Figure: Scan performed in 2010. (a) MXR scanner, (b) color image of print, (c) quality image, (d) metallic surface, (e) depth encoded and (f) gray level encoded depth. The print was barely visible with the naked eye but by using these scans to aid in inspection, various features became apparent under visual inspection.

Figure: Scans from 2006 with bare footprint as control. (a) Bottom of boot gray level depth encoded, (b) encoded print in mud, (c) encoded print in mud under 1 inch of water, (d) encoded control footprint and (e) photograph of boot print in mud.

PHASE 4: (December 2009) Mixed Resolution: Historical Cemetery in Lexington Kentucky.

While waiting for the data processing to be completed for the RAHAS scanning system, we directed our attention to remote mixed resolution scanning. By combining technologies from the previous 3 phases, Dr. Hassebrook constructed a battery operated Mixed Resolution scanner shown below:

The unit above can scan at any angle and is battery operated. It requires a tethered laptop to operate the system but gives immediate feedback to the operators. Chris Begley, Laurence Hassebrook and Walter Lundby tested the unit out on a historical site in Lexington Kentucky. Unfortunately, we made a mistake in the high resolution scanning so the first attempt did not allow for mapping the color images. We show them separate below along with the depth imagery. It is important to see what the headstones say so we used two levels of depth enhancement, surface normal rendering which basically treats the surface like a metallic one and depth encoding that we developed for visualizing 3-D fingerprint ridges. The data is ordered in easiest to most difficult, in reading the text.

Above: One of the most historical stones. Sister Ann Spalding died in 1848. It should be noted that our team thought that the date was 1818 until we performed the scan. The thin front line of the four along with the angle of the primary groove differentiate the “4” from the “1”.

Above: A vertical stone dating to 1861 for Mary Ann.

Above: Horizontal stone. Note the far right where we viewed the relief from the back side so the letters stand out from the surface. This way, the least worn part of the text is the most exposed in 3-D space.

Above: The most difficult stone that we found is shown above. It starts out with “In Memory of” and gets more difficult to read from there. The problem with this one is there are grooves similar to that of lettering that clutters the stone. As you look at the renderings you may start to see the underlying text. Our conjecture is that this text is salvageable.

PHASE 3: (July 2009) Rotate and Hold and Scan System: Honduras Expedition

In July 2009, Chris Begley headed up an expedition to capture 3-D scans of archeological features in the Jungles of Honduras. Eli Crane accompanied him with a prototype technology representing a radically new concept in 3-Dimensional scanning. We refer to this scanner as the Rotate and Hold and Scan (RAHAS) scanner. It achieves a non-ambiguous depth scan combined with a high resolution texture map. The device operates on the processors in conventional digital cameras and power efficient pattern projection system. There is NO LAPTOP necessary. The quality of the data is accessed through the digital camera playbacks and the 3-Dimensional information is fully processed upon return to the home base. We are presently developing the processing capability for this project and designing the next generation of RAHAS scanner systems.

Above: (left to right) Chris Begley and Eli Crane. (center) Head unit of RAHAS system and (right) Eli setting up for a scan.

Components of the RAHAS scanner.

Above: Data processing algorithm flow of RAHAS. The depth processing flow starts upper left and proceeds counter clockwise. The high resolution color texture image is in upper right corner. The Rotate and Hold starts in the epipolar alignment so that the snakes or stripes can be non-ambiguously identified. These snakes are then tracked through the rotation and result in a non-ambiguous set of snake boundaries. The “and Scan” process is conventional PMP and yields the wrapped phase image. Combining the snake boundaries with the wrapped phase results in a non-ambiguous unwrapped phase which is then transformed to world coordinates. The color texture is then mapped based on a prior calibration process to the same coordinate frame as the depth world coordinates. The depth information is obtained from a video sequence which always lags the digital camera image in resolution. Thus, the depth is interpolated up to the color texture resolution. Most scanning systems don’t support mixed resolution so are always lower resolution than the latest digital camera. The RAHAS method resolves the problem and can be used in conjunction with any digital camera technology.

PHASE 2: (May 2009) “HatBox” 3-D scanner

In May 2009 we constructed the a portable 3-D scanner we refer to as the “hatbox” scanner because of its cylindrical shape. This unit had a battery option and requires a laptop computer. It is similar technology to the CSS Alabama hardware but uses a low power LED based projector that can be run off batteries. The video camera is powered by the laptop battery.

Above: The hatbox scanner is operated by a laptop. The camera can be operated by the laptop battery and the projector is battery powered when AC is not available.

Above, the unit has a camera and LED projector. It reassembles into a compact hatbox form factor.

PHASE 1: (November 2008) Non-Ambiguous Mixed resolution 3-D scanning

In November of 2008 we assembled a scanner system by combining a 12 megapixel digital Canon G9 camera to capture a color texture image of an artifact from the CSS Alabama. The depth was capture by a 2 megapixel B&W digital video camera in conjunction with a commercial projector. The depth is non-ambiguous which means that all points have an absolute coordinate over a large depth range. The lower resolution depth was upsampled to match the color texture image. Thus, in addition to having a conventional photograph, we obtained a complete 3-Dimensional interpretation of the artifact. The keystone of the system is software, not hardware. We can combine any commercial cameras and projectors without special hardware.

Above: (left) CSS artifact, (center) structured light projection and (right) merged depth scans.

Above: Mixed resolutions of color texture (66.4 micron pixel spacing) and depth (136.9 micron pixel spacing).

Scanner setup and components for artifact scans.