A 1.219Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE
It seems that every surveying magazine now overflows with stunning 3D images of everything from people, to cars, to historic landmarks, to infrastructure. Point clouds of intricate structures and shapes naturally captivate the geometrically adept mind.
But what about the practicality of such a colossal amount of data? How do you get from the obscene quantity of points to something useful for a paying customer? Considering the overall ticket price of laser scanning hardware, software, training and computer systems, cost becomes an important consideration for the potential buyer.
This is where Maptek steps in. Maptek manufactures I-Site hardware in Australia and markets it and I-Site software through its global network of 12 offices. The company began with the development of mining design software in the early 1980s. In the late 1990s I-Site software was developed in order to take laser scan data and make useful modeling data for use in their mining application software. Unlike other companies that produced hardware first and later developed software to make use of the data, I-Site did it the other way around, developing software first, and later manufacturing their own scanner, the 4400.
A Dirt Gun
The 4400 is built for rough modeling in adverse conditions. It is spec’d to have a distance precision of 5cm, an angular accuracy of 0.04 degrees, and an impressive range of 700 meters (LR models) to reflective surfaces, 150 meters to black coal or 600m to rock or concrete.
I met with Maptek personnel John Dolan and Scott Schiele to try out the equipment. In the field, operation is very similar to using a total station. The user sets up a tripod and tribrach with optical plummet over a point. After placing the unit on the tribrach, the user measures the height of the instrument and directs the scope to a backsight target, which in our case was a prism pole with a bipod. Instead of using tangent screws, the user directs the instrument to the backsight using an infrared controller that reminded me of the remote keyless entry fob for my truck. Once collimated on the backsight, the user controls the scanning operation from a ruggedized tablet PC that is linked to the instrument using an Ethernet cable. If no control points are available from which to observe, I-Site Studio also allows several methods of registering scans (i.e., stitching them together).
Once set up and oriented, the user initiates a quick preliminary scan. This scan initiates the electronic level compensator and uses the 37 megapixel camera to generate a 360-degree by 80-degree panoramic view of the site. The user views this photo scan and can make adjustments to the exposure for bright sun or cloudy/low light conditions. The user can select either the high resolution or medium resolution scan setting depending on conditions. A 360-degree scan takes only six minutes for a standard resolution capture and about 12 minutes for high resolution. Once captured, the scan is sent to the tablet PC for storage. A standard resolution scan will take half the amount of time to download the data from the tablet compared to a high resolution scan. As the scan is being performed, the digital camera is also collecting photo pixels and each scan point is registered to the photo scan allowing for some powerful visual interpretation back at the office. If the user doesn’t need a full 360-degree scan, he may simply window the horizontal extents of the desired area on the panoramic 360-degree photo to limit the scan. The vertical extents (80 degrees) will always be scanned regardless of user setting.
Physically, the 4400 is a brute beast. There is very little plastic on the scanner, which tips the scales at about 30 pounds, including the onboard removable battery, giving it a very rugged exterior. It is encased in a solid metal housing.
For our testing, I took John to Nettleton Station, a small tank farm located fifteen minutes from the office. Each tank is surrounded by a large earth wall or is placed in a depression in the earth to contain accidental spills in the event of a tank or plumbing failure or fire. The volume of the pit is designed to contain whatever volume the tank capacity is able to hold. In order to verify this condition, surveys are performed from time to time to determine the volume of the fire wall or pit. It was with this scenario in mind that we set out to scan one of the tanks at Nettleton.
As John and I scanned our tank site, we carried the unit outside of the Pelican case he brought it in, setting it on the ground at each set-up until the tripod was ready. John set the unit up on the first observation point and turned me loose on the remaining three. There were only a few basic steps in making the set-up, initializing the unit and commencing the scan. Once a scan was completed, it was sent via the Ethernet cable to the Tablet, where it was compressed and stored. Nothing was left in the 4400 itself.
Field training was very quick. By the third occupation I was able to perform the scan from start to finish by myself. From my limited time using the equipment, I came away feeling that while field staff should be meticulous, they need not be highly technical, just trainable. Depending on the types of surfaces being scanned, they will need to be able to visualize what is visible to the scanner and what is not in order to maintain necessary coverage. I was very impressed with the ease of use of the system in the field, and within an hour and a half we completed our scanning and were ready to return to the office.
3D at Warp Speed
Back at the office, John copied the files from the Tablet to a USB memory device and from there onto my computer. My present computer is a single core processor running 2.4 GHz. As I thought about the sheer volume of points I was about to work with I got nervous. Just trying to use my CAD program with aerial photos and a threeor four-thousand point file gets to be slow. Was I in for a surprise! I’m still not sure how I-Site did it, but somehow they managed to utilize a graphics engine that never seemed to get tripped up.
After John explained the mouse buttons, I was able to pan, zoom, rotate and change central rotation point very quickly. The hundreds of thousands of points in the scans zipped across the screen effortlessly with none of the regenerating blackouts one might expect in CAD. Files opened very quickly. It seems we can often get hung up on hardware speed, but I was reminded once again how much positive effect good programming has on overall system speed.
Imagine, if you will, a craftsman building an intricate widget at a work table. He’s got his tools laid neatly in front of him, with the components arranged to his left. As he works, he takes a component from his left and incorporates it into the widget. As he’s working, he may even set an assembly to the side and work on another assembly until he’s ready to put it all together. This is perhaps the best way I can relate to you the I-Site Studio workflow.
Once the scans are imported into the current working file, their file names are visible on the left side of the screen. From there, they may be dragged-and-dropped into a working pane where they can be visualized in 3D. One by one, each scan that is to be worked with can be dropped in, or the user can work one at a time. If you are working with control, a simple text file with Easting, Northing, Elevation and Point Number are all you need. In the field, the scan is named based on the occupied point number, the backsight point number, the range setting (long or short), the resolution (high or medium) and whether the unit was leveled or not. Thus, automatically, once the survey data is imported, the scans jump to their proper points in space.
If survey data was not used, the scans can be registered by manually selecting various common points in the scan or by allowing the software to search for apparent common points between scans. These processes also allow for statistical analysis of how well the scans were put together. Regardless of which method you use, the process is very fast, however, using survey data was by far the fastest and provided the best results.
Once the scans are registered there is an incredible amount of data that is either undesirable or redundant for the project. Also, there are likely to be occasional spikes, false returns, or returns from a bird or car or some other transient object. I-Site allows users to quickly remove unwanted data by manually selecting pixels, or by filtering the data using some ingenious options.
For instance, you likely don’t want to include a tall clump of grass in your model, so I-Site allows you to remove all but the lowest laser point within a userspecified grid. Therefore, you can set it up to remove everything within a 4ft. x 4 ft. grid except for the lowest point in that grid. Suddenly you’ve removed a lot of shots that are very densely spaced and removed any high zingers that don’t belong. The data itself isn’t erased and can be brought back anytime, it is just removed from the work space.
Visualization is important, of course. Bringing in four or five scans as we did gets very busy. To help distinguish between them, the user can change the color of each scan. If a cluster of points doesn’t look right, it isn’t very difficult to determine which occupation they originated from. The colors can also be set to the pixel color from the 37 megapixel photo scan that was observed in the field. Each point now takes on the color observed in the field offering some impressive views.
With our points neatly stitched together, the next step was to build a triangulated surface file. The triangulation file options allow the user to develop the surface from a topographic viewpoint or spherical point. Using the spherical perspective allows the user to model the surface under outcrops as they actually are instead of connecting the surface lines from the top of the shelf to the ground below.
Once the surface was generated, we were able perform volumetric calculations and create contours (Figure 1). Recalling that our hypothetical job was to determine the volume of the pit, we needed to determine at what point the wall would be breached by a rising pool (Figure 2). In order to quickly accomplish this task I created a simple four corner planar surface. I-Site allowed me to modify the elevation of this plane and as you can see in the Figure 2, I slowly "filled" the pit by increasing the plane’s elevation. Once I found the overflow line, I trimmed the pit surface at this level. From here I calculated the volume between the basin and the plane (Figure 3).
According to I-Site, there were on the order of 65,000 individual facets in the calculation. This required a grand total of 4.4 seconds to determine, again extolling the efficient programming of the I-Site team with a grand total of 364,000 cubic feet of volume.
The I-Site 4400 laser scanner is a solid unit having performed in some of the most extreme environments on the planet, and while not necessarily designed, say, for architectural scanning, it is right at home getting modeling terrain in the "real world" of dust, mud, hot or cold, wet or dry. When combined with the impressively fast I-Site Studio, getting to a real marketable product is fast and easy. With the ability to directly export surfaces and contours as AutoCAD version 2007 DWG files as well as several versions of DXF files, shuffling final products into existing design networks is uncomplicated. If your operation needs stock quantities yesterday, you should definitely consider I-Site’s turnkey system.
Shawn Billings is a licensed land surveyor in East Texas and works for Billings Surveying and Mapping Company, which was established in 1983 by his father, J. D. Billings. Together they perform surveys for boundary retracement, sewer and water infrastructure routes, and land development.
A 1.219Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE