If youve been scanning for any length of time, youre bound to come across some pretty interesting things. This was the case for my team not too long ago.
As a service provider focused primarily on building documentation, we receive a lot of requests to perform above ceiling plenum surveys for the purpose of creating an as-built model of complex SMEP (Structural, Mechanical, Electrical and Plumbing) systems. Scanning above ceiling can be challenging for many reasons, one of which is targeting.
Historically, weve found flat magnetic paddle target plates have come in handy when doing above ceiling work. There never seems to be a shortage of metallic objects that we can attach them to. However, with so many obstructions (ducts, pipes, framing, etc.), it can be challenging to ensure lines-of-sight are maintained between the targets and the scanner when placing our targets above the ceiling. In addition to line-of-sight issues, weve found it can really slow down our workflow to constantly be climbing up and down ladders to spin and adjust the targets each time the scanner is repositioned. It was this frustration that let us to consider testing out spherical targets.
We had heard stories of people who have made their own spherical targets out of things like balls, globe lights, etc., but we didnt feel comfortable with this, so we ordered five SECO Omnidirectional scanner targets. We figured it was too important to take a chance on a homemade solution and that it would be worth the added cost to ensure we were using an object that was designed specifically for our intended use as a laser scanner target.
Online ad for the SECO spheres we found
When our targets arrived we decided to test them out in our office. We used our Leica HDS6000 phase based scanner. It was immediately upon viewing our first scan that we noticed something was wrong. The target had shifted. Just the target? Not the pole, and apparently nothing else in the scene? Hmmm??? The sphere was the correct size, but it was just shifted.
Image from a single scan
So we decided to run a couple more scans and register them together. Wow, bizarre! We confirmed that the pole registered together nicely and was not shifted. Again, why just the sphere?
The image of the registered sphere reminded me of playing with bubbles as a kid.
O.K., so this was clearly a problem. It just didnt make any sense. So we decided to do some further testing. Our theory was that it had to do with the material the sphere was made out of. Could the laser light actually be penetrating into the surface of the sphere causing the distance calculation and the resultant placement of the points to shift further away from the scanner?
We decided to cover the surface of the sphere with blue painters tape and re-scan. The resulting scan showed the sphere in its proper place. In our minds this proved it had to do with the reflectivity of the material the sphere was made out of. In the mean time I decided to reach out to my friends and colleagues to see if anyone else had experienced anything like this. Here were some of the responses I received
That is strange; my only suggestions would be to return the spheres to Seco with the data for a refund and continue to scan with our 6" tilt & turn targets and cloud to cloud.
The only thing I have seen like this was with Leicas loaner HDS6000 I had in DC last winter. It would do something similar when acquiring targets. my recommendation would be to drop the scanner (in its case of course) one or two times and re-test.
I know a few people who have used them and abandoned them. Too many issues like what you are experiencing.
From what I understand, when modeling the surface for the sphere to fit intoany anomalies or noise can cause the problems you are having. This is much more pronounced in a sphere than a flat surface, such as a paddle. If you notice when you scan a plain curved surface all sorts of scatter can occur. Now on a sphere it is happening in all directions.
When I used the spheres with the Trimble scanner, when I did the registration, I wound up with egg or football shapes when the software tried matching the spheres. We wound up switching to adhesive targets before switching to Leica equipment. I did have to do a lot of cleaning of the cloud around each sphere but I never had the issues you are having with the 2 images you showed.
Very odd, cant say Ive ever seen that before
As it turns out, one of our guys has had this problem before. Apparently, he recognized it very early on and switched to flat targets to avoid the issue. He wont use spheres anymore and thinks that it has to do with their color. Others here think that its due to a software error in calculating the centroid of the sphere. Either way, we use flat targets exclusively now.
Ive spoken to several people here and at other companies to see if they have any idea why that may be occurring. They didnt really have any other possible scenario other than what youve probably surmised. I dont think its a SECO issue, in my opinion it is definitely a Leica/Trimble/Cyclone issue.
Wow, thats weird. Weve never encountered that.
The only info I could find about spherical targets maybe having trouble was based on their outer coating not being sufficient to keep the laser light from penetrating the semi-transparent plastic sphere, thus allowing the laser light to bounce around in the sphere and confuse the scanner.
We do not use spheres in our work but have had similar occurrences with targets. Our targets are standard Leica scan targets and are in two colors, blue and White. The blue section of our targets appears closer than the white section during acquisition. Our software accounts for this when it models the target so it doesnt’ affect us, but it does prove that return intensity makes a difference.
It is not uncommon for scanners to have range bias as a function of the material properties. The degree of the bias is pretty high for the data shown (based on the tripod in the scene).
Scanners are generally calibrated to work accurately with a range of return intensities, but both phase and time of flight will become less accurate for highly reflective materials. This characteristic is analyzed extensively in the University of Mainz reports that analyze various scanners. The planar targets that we make have a highly reflective blue part and a normal white part. The highly reflective blue part is only used to identify potential targets in the scene (since the intensity return values are higher than surrounding points) and are not used to fit, since they are less accurate. The points on the white part are used to fit the center of the target. So, the target can be found easily and fit accurately. The problem with the spherical target offset using 6000 data described below is odd, since that scanner does very well on accuracy over the normal range of intensity return values. If the sphere was highly reflective then coloring it something more neutral would improve results, but if it wasn’t highly reflective it may be that the scanner needs calibration.
I would suspect either a hardware sensing error due to glare or unwanted reflection, or possibly even a software glitch.
If the surface of the sphere has a high shine to it can appear like the points are actually penetrating the surface. I have never seen it with spheres but have seen it with polished marble surfaces.
For our final test we used a Disto. We took a shot from the sphere to a wall and got a reading of 2.00 feet (as we expected).
We then took a shot from the wall to the sphere. The result this time was different. 2.06 feet. This seemed to confirm our theory that the laser was penetrating into the surface of the sphere.
My friend from Leica had this to say
I have talked to a few people in regards to this issue and it is a common problem with the SECO spheres. These spheres are not an approved target from Leica and the problem being is the laser is able to penetrate the white plastic and gives a false reading, which supports your test when you put tape over the sphere it was able to give you a good reading. A quick fix is to spray paint the spheres with flat grey primer type paint.
In addition, we got the following response back from the dealer that sold us the spheres who had contacted his SECO rep
He said he [SECO Rep] was aware of this problem, and that it is only an issue with certain scanners (apparently your Leica 6000 being one of them). The problem is that certain scanners emit a more powerful beam than others and that this beam is actually penetrating the surface of the sphere, giving you the strange data capture.
One other important thing to note is that this particular brand of sphere is made to glow in the dark – a great idea to help locate your targets when scanning at night. However, as we can see this material is designed to absorb light and is clearly not appropriate as a target for laser scanners.
Based on all of the feedback we received and the testing we performed we decided to paint the spheres and re-test. We were satisfied with the results we got after doing this.