| Re: Compute area between curves [message #62921 is a reply to message #62821] |
Thu, 16 October 2008 05:18   |
Jeremy Bailin
Messages: 618
Registered: April 2008
|
Senior Member |
|
|
On Oct 15, 11:42 pm, Craig Markwardt
<craigm...@REMOVEcow.physics.wisc.edu> wrote:
> frankosuna <frankos...@gmail.com> writes:
>
>> I need to come up with some measurement to see how close the edge
>> detected ring from the image and the ring from the wireframe are. They
>> are ellipses and they do cross since we're trying to get them as close
>> as possible.
>
> OK, it still sounds like the "best" way to do it is take your points
> and re-sample them to the same grid, so they are directly comparable.
> In principle you can re-sample in X, but I think it would be better to
> convert the original points to R,PHI polar coordinates and sample to a
> uniform PHI grid. At that stage, you can compute
> TOTAL((R_MEAS - R_WIREFRAME)^2)
> which is effectively a chi-square, which then also measures goodness
> of your fit. [ I.e. forget about integrals. ]
>
> Craig
>
> --
> ------------------------------------------------------------ --------------
> Craig B. Markwardt, Ph.D. EMAIL: cbmarkwardt+use...@gmail.com
> ------------------------------------------------------------ --------------
If the difference is just translational, I'd still just use an FFT to
cross-correlate the original images and find the peak...
-Jeremy.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
