Dear DAVID Community,

As you are probably aware, the entire DAVID company, team and product range has been integrated into HP Inc.
This forum is now read-only. Relevant content has been migrated to HP and merged into HP's Support Forums on November 1, 2016.

To start new discussions for 3D Scanning, please register and post your new topic at the HP Support Forums

Camera Calibration Maths

The place for questions, problems, comments and tips regarding the camera calibration.

Camera Calibration Maths

Postby ben2008 » Mon Feb 11, 2008 5:29 am

The scanning I want to do requires the laser and camera to move along an long object so I cant directly use David.

I was wondering how it is that David takes the (X,Y) information of the circles on the calibration screen and uses the right angle geometry to convert the points into the (x,y,z) information. I imagine it makes use vector dot products and vector lengths but i cant figure out how to simplify my equations. Any ideas/links?
ben2008
 
Posts: 9
Joined: Sun Jan 27, 2008 7:51 am

Postby MagWeb » Mon Feb 11, 2008 7:46 am

Hi Ben,

just an idea to use DAVID directly for your project:

Use a fixed windowsetup like Hal described at:

http://www.david-laserscanner.com/wiki/wax_head_by_hal

and turn the backgroundfilter to value -1 (so the filtered areas don´ cut the object).

To do this, you need the Version of DAVID Sven posted at:

http://www.david-laserscanner.com/forum/viewtopic.php?t=351

This way you are able to scan parts of any kind of object and put it together with shapefusion finally.

Maybe that works

Gunter
MagWeb
Moderator
 
Posts: 2499
Joined: Wed Jul 18, 2007 8:48 pm

Postby Bongobat » Mon Feb 11, 2008 3:58 pm

Hi,

I dont mean to sidetrack your post here but I have some related questions. I made a 3d mock up of Mattia's (Hal) scanner with some modifications. I have posted it here.

For Mattia: Is your orginal setup scalable? (Do the panels slide on the cross)

For Anyone: If you have the panels slide on the the T frame could'nt you calibrate in a closed position then open up (equal amount from center) to an arbitrary point. You would still be scanning in a corner and still 90 degrees just a bigger one. Do you do this Mattia?
Attachments
halscanner.pdf
Hal's Scanner
(33.22 KiB) Downloaded 430 times
DealExtreme red line laser, Microsoft Lifecam HD-5000
User avatar
Bongobat
Moderator
 
Posts: 1002
Joined: Mon Apr 02, 2007 10:02 am
Location: Ohio, USA

yes

Postby hal » Mon Feb 11, 2008 5:02 pm

Ciao, Bongobat, ciao, MagWeb,
Yes, Bongobat has centered the goal. His 3D model explain well the my cross. Yes, this setup is scalable: the panels slide on the cross, on the short side, and the camera slide on the long side.
I don't open the panels after calibration: I move the panels, only after I put in the scene the pre-made rigid calib-angle, I do the calibration step and I romove the calib-angle, like visible in my old tutorial at wiki link:
http://www.david-laserscanner.com/wiki/wax_head_by_hal
, first images in "Hardware Setup" chapter.
This is an old solution: I had need a light (little heavy) structure, portable. This old cross was just a little inaccurate. Now I'm working on a more precise one (Not exactly a cross, but two panels on one binary).
My goal is a structure such as MagWeb's structure, but lighter and portable. Soon as possible info and images.
PC: DELL Alienware 17, Intel I7-4710MQ, 32 Gb Ram, AMD Radeon R9 M290X 4GB.
SO: Win 8.1 64Bit.
CAMs: USB 3.0 CMOS B/N (1280x960, 60 fps).
LENS: 12 mm.
PROJECTOR: DAVID SLS1 ACER K11.
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Postby MagWeb » Mon Feb 11, 2008 5:50 pm

Hi Greg and Mattia,

it is possible to calibrate closed and open the window afterwords:

You make DAVID think it's the same corner. So DAVID simply detects the object smaler for the corner "moved back".
You have to switch of the background-filter for it would hit the objects (as said before).
(whether the lens distortion is set right this way - I don`t know)

I dropped my test in this direction, for the movement of the wings has to be very accurate and symetric to avoid destorted results.(Greg: I think you should add a upper slidegide) I prefer to calculate the new position of patterns fixed of the wings with my calibcalc and get better results this way. But that works too.


Gunter
MagWeb
Moderator
 
Posts: 2499
Joined: Wed Jul 18, 2007 8:48 pm

Postby Bongobat » Mon Feb 11, 2008 6:39 pm

hi everyone,

Yeah the camera calibration was what I was wondering mostly about. My goal now is for a setup like Mattias but perhaps constructed of metal or plastic pipe. For the panels i was thinking very thick (1 inch or more) styrofoam. I had some really thick formed packing material that seems very straight and true. Laid glass on it and the gap between only vaires by the thickness of paper. The pieces I have are too small to test. Average panels size around 30 cm square glued to a bearing. I was thinking with the small size weight of the panels plus tight tolerance in pipe and bearing I might get satisfactory results. Im teaching myself how to sculpt and I want it for mainly small items but would like the possibility of a more scalable solution.

P.S The foam would be painted with magnetic paint. My calibration printouts are on magnetic sheeting. So with panels in closed position I just have to place the printout on the foam calibrate then remove printout, then open up the panels to scan.
DealExtreme red line laser, Microsoft Lifecam HD-5000
User avatar
Bongobat
Moderator
 
Posts: 1002
Joined: Mon Apr 02, 2007 10:02 am
Location: Ohio, USA

Postby MagWeb » Mon Feb 11, 2008 7:07 pm

Hi, me again,

What´s the disadvantage of calculating the origin? Put your patterns on the planes, put in the windowwidth to the calculator (you could place a scale at your T), store, open DAVID and calibrate. This way you get right data without tricking DAVID. I allways scan that way (nearly never with closed corner).
Note - Calibrating on a closed corner and scanning through a window: Scanns of the same object, taken through different windowwidths will never (without scaleing) fit together for they own different sizes.

Gunter
MagWeb
Moderator
 
Posts: 2499
Joined: Wed Jul 18, 2007 8:48 pm

Postby Bongobat » Mon Feb 11, 2008 7:27 pm

Yes Gunter perfect! You had the answer all the time but somehow I overlooked it. http://www.david-laserscanner.com/forum ... ight=#1774 Great tool! Thank you.

Edit: I redesigned my idea. I added stability rail but would prefer not if I can figure out how to get away from it. Removable turntable. Is this like your setup Gunter?
Attachments
newscanner.pdf
Redesign
(48.58 KiB) Downloaded 452 times
DealExtreme red line laser, Microsoft Lifecam HD-5000
User avatar
Bongobat
Moderator
 
Posts: 1002
Joined: Mon Apr 02, 2007 10:02 am
Location: Ohio, USA

Postby ben2008 » Tue Feb 19, 2008 1:25 pm

Hi Gunter,
Thanks for the quick reply, unfortunately for my setup I need to have both the camera position and laser fixed relative to the background and to then have the rig move along the object.

I have attached my current attempt to solve for the z coordinates of the background points. My idea was to use 4 points on the background and create 4 equations from Equation 8 using the points

1 2
3 4

to make the vectors 12, 13, 24, 34 and then use the dot product of the vertical ones which in theory should create 4 unique equations with 4 unknowns which could be solved with matrix manipulation. Can anyone see where my mistake is in getting to equation 8?
Last edited by ben2008 on Sat Feb 23, 2008 2:51 am, edited 1 time in total.
ben2008
 
Posts: 9
Joined: Sun Jan 27, 2008 7:51 am

Postby ben2008 » Thu Feb 21, 2008 10:22 am

doh! len ij == len ik != len jk

*sigh*

0 does equal 0

back to the drawing board
ben2008
 
Posts: 9
Joined: Sun Jan 27, 2008 7:51 am


Return to Camera Calibration

Who is online

Users browsing this forum: No registered users and 1 guest

cron