[argyllcms] Re: Fwd: Simple how-to on camera profiling
- From: "Guy K. Kloss" <g.kloss@xxxxxxxxxxxx>
- To: "argyllcms@xxxxxxxxxxxxx" <argyllcms@xxxxxxxxxxxxx>
- Date: Sat, 20 Mar 2010 10:32:36 +1300
On Sat, 20 Mar 2010 06:20:55 Ben Goren wrote:
> On 2010 Mar 19, at 9:45 AM, Karljohan Lundin Palmerius wrote:
> > May I ask how you do to find the coordinates of the fiducial points?
>
> Actually, I cheat. I open the file in Photoshop and crop it so only the
> target is included. I might rotate it in Photoshop as well.
Well, that doesn't fully work. As soon as you've got some perspective
distortions in it that approach will fail. That was also the reason why Graeme
has introduced the "fourth fiducial point" to be able to overlay a the mesh in
these perspective cases.
For finding the coordinates ... yes, that's a bit troublesome. In the past
I've used different approaches. Initially by reading the coordinates after
opening the image in Gimp (I bet PhotoShop will be just as usable, but for a
lot higher price). Then I've just mis-used a little Python application (usint
the ctypes-opencv bindings) that I have once written for another project. With
it I could easily record rectangular regions of interest (ROIs) for an
application that I've built for industrial quality control. It would
additionally spit out the coordinates on the terminal, from where then I have
copy/pasted them into a shell or Python script for mass application of scanin.
Unfortunately the camera lacked in resolution, so tolerances of a few pixels
mattered, and often I had to hand tune the *exact* position of the fiducial
points after visually inspecting the debug output TIFF that scanin produced.
HTH,
Guy
--
Guy K. Kloss
Institute of Information and Mathematical Sciences
Te Kura Pūtaiao o Mōhiohio me Pāngarau
Massey University, Albany (North Shore City, Auckland)
473 State Highway 17, Gate 1, Mailroom, Quad B Building
voice: +64 9 414-0800 ext. 9266 fax: +64 9 441-8181
G.Kloss@xxxxxxxxxxxx http://www.massey.ac.nz/~gkloss
Other related posts:
