Re: zernike moments question

Hi Adam

I used Zernike moments about 10 years ago as a way to compactly
describe images of micrometeor impact crators. I think their history
comes from describing optical distortions and they are only used as an
orthogonal basis set when used within that circular domain. So,
whilst the functions have values outside x^2+y^2<=1, the set of
Orthogonal Zernike moments are only defined inside this range. If you
are projecting your images onto Zernike moments then you will need to
only use pixels within x^2+y^2<=1.

You will need to scale your pixel coords so that you use this range
with the functions. Divide your pixel coords by the image width and
make sure to add 0.5 to center the functions.

I think I had to use a black mask with radius of R=W/2 to black out
the corners of the images too, since they lie outside the range.

One other thing, i found Zernike moments quite sensitive to the
coordinate used for central position, so I gave up with them as robust
statistics in the end.

Regards

Jason
www.visionexperts.co.uk

Adam Chapman wrote:
Hi,

I have to calculate the zernike moments (http://homepages.inf.ed.ac.uk/
rbf/CVonline/LOCAL_COPIES/SHUTLER/node1.html) of an image as a step in
an image processing algorithm.

The problem I have is that you are only supposed to make the zernike
moment from the region where x^2+y^2<=1. This suggests that I have to
compress my pixel grid to fit in a circular shape with an edge radius
of 1.

Am i correct?

Adam
.