PanoTools mailing list archive

Sender:Thomas Niemann
Date/Time:Fri, 26 Aug 2005 09:41:06 -0700
Subject:new calibration procedure


Rik Littlefield and I have been working on a new calibration procedure for 
lenses. Currently in the experimental stage, I thought I'd share with you 
what we have so far. This technique reduces the number of points I plot from 
20 to 8. It also reduces the requirement that the target be parallel to the 

Instead of using t1/t2 points for horizontal/vertical alignment, we're using 
t3 and t4 points. In particular, two points defining a long t3 feature are 
specified, then two additional points are placed within the feature. Same 
for t4. The order of placement is important (long before short) as the 
algorithm might not work properly otherwise. See the following illustration.

Points (t3a, t2d), (t3b, t3c) are defined. Point t3c is placed at the point 
of maximum distortion. Point t3b determines the rate of curvature. A similar 
sets of points are defined at t4. This is sufficient for an a-b calibration 
on a non-fisheye lens. For calibration of just one coefficient (b-only), 
just a t3 line is required.

I just finished calibrating a telephoto converter for a canon S2 using 
b-only coefficients. Since the S2 is a zoom lens, I had several images to 
calibrate. At 4 points/image, I was done in no time and the results were 

I recalibrated the Sigma 18-125mm lens. This one has a strange hook to 
distortion in the corners. For this lens I added 2 points (for a total of 
10), and the calibration was better than the original calibration that used 
20 horizontal/vertical points.

Special thanks to Rik for explaining the code. We're still exploring other 
options, and these results should be viewed as experimental. If you have an 
image that doesn't calibrate well with this procedure, please contact me. 

------------------------ Yahoo! Groups Sponsor --------------------~--> 
Fair play? Video games influencing politics. Click and talk back!

Yahoo! Groups Links

<*> To visit your group on the web, go to:

<*> To unsubscribe from this group, send an email to:

<*> Your use of Yahoo! Groups is subject to:

Next message:

Previous message:

back to search page