Maths interpolation


Thread Starter

J-F Portala

Hi listers,
I am working in the filed of
industrial image processing, and I have the following problem. On a camera, the lense creates image distorsion on the edge of the camera
field. I would like to compensate it mathematically. I intend to use a sheet of paper with a black grid. I know exactly the location (in mm) of the intersection points on
the grid, and I can measure on the image, the variation on the intersection points (in pixels).
I can compensate the pixels corresponding to the intersection points, but for all the pixels between 2 intersection points, I need to
interpolate the values.
My problem is that I don't know what kind of interpolation I can use in 2 dimensions.

Thank you for your help.

J-F Portala
SoViLor company
[email protected]
You have linear interpolation which is the easiest, logarithmic, exponential...
It all depends on how the data that you're
trying to compensate for is distorted.
If you can represent your compensation an
equation, or better yet a differential equation, then you can calculate the coefficients or constants that can be plugged into standard interpolation formulas to get you going.
Ready to run tools that provide good libaries are labview, Labwindows both from national instrument. Mathlab from mathworks. Even excel can be easily programmed to process the data.
Go to the libary and consult any good book on numerical analysis if possible.


Alberto Maldonado

Mr. Portala:

I think You have 2 cordinates (x,y) if you have distortion in "y" You can only interpolate in "y" and not in x, but if You have distortion in x too,
You can use the same method from "y" coordinate for "x".


Eng. Alberto Maldonado-
E-Mail: [email protected]
Fax: 591-4-294171
Phone 591-4-294171