Re-projected interest points are off

[GoogleCodeExporter]

First of all, thanks for the great toolbox. 

I have one problem though. Corner detection seems to have some strange error. 

I've attached a view of the erroneous output, and the training set. 

Original issue reported on code.google.com by the.tyle...@gmail.com on 22 Nov 2010 at 4:02

Attachments:

• [erroneous output.jpg](https://storage.googleapis.com/google-code-attachments/procamcalib/issue-9/comment-0/erroneous output.jpg)
• pics.zip

[GoogleCodeExporter]

The corner detection is performed via the Bouguet toolbox...
Maybe it's because of the parameters (number of squares, size, etc...)... 
usually the corner detection works well. 

Original comment by david.f...@gmail.com on 22 Nov 2010 at 3:23

[GoogleCodeExporter]

probably the corner size is too big, try the smaller size, then the corner 
detection is more accurate.

Original comment by manches...@gmail.com on 26 Oct 2012 at 8:28

[GoogleCodeExporter]

Ignore the above comments, they are BS.

yeah this toolbox is fairly crappy and buggy as hell. That problem is because 
the initial guess of fc, cc and kc is waaaay off. In cam_proj_3d_points, change:
ProjectedGrid_2dpoints_cameraPlane = extract_grid(I,wintx,winty,fc,cc,kc,dX,dY);
TO
ProjectedGrid_2dpoints_cameraPlane = extract_grid(I,wintx,winty);

And it works. But then still you probably get the 'ill conditioned' problem as 
everyone gets. i doubt if anyone ever *really* got this toolbox working, 
everyone who says "great toolbox" actually has questions that indicate they do 
not know what they are doing. This toolbox should be elaborated with test 
images to show that it actually works. Its not really a toolbox otherwise.

Original comment by timbo...@gmail.com on 18 Nov 2012 at 5:12