[Insight-users] Re: GradientDifferenceImageToImageMetric?

[Vicente Grau]

Damiaan,

are you using the CVS version? Some time ago, I reported a bug on this 
metric that was then corrected. I didn't use the gradient calculation 
function, so I can't tell if it works well or not. Anyway, if I'm not 
wrong, I think the delta value is used here just to approximate the 
derivative with a finite difference calculation. In other metrics, the 
derivative is calculated explicitly.
It doesn't seem that many people have used this metric in the past.

Vicente

>------------------------------
>
>Message: 3
>Date: Thu, 16 Feb 2006 22:48:08 -0500 (EST)
>From: "D.F.Habets" <dhabets at imaging.robarts.ca>
>Subject: [Insight-users] GradientDifferenceImageToImageMetric?
>To: insight-users at itk.org
>Message-ID:
>	<Pine.GSO.4.60.0602162241580.24936 at theremin.irus.robarts.ca>
>Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
>
>Hello everyone,
>
>    I am trying to use the itkGradientDifferenceImageToImageMetric, but 
>cannot get it to return anything but zero values for the Derivative. I 
>looked through the list archive and found some references to bugs that 
>were fixed in it, but could it be it's still problematic in the latest 
>stable release?
>
>I looked at the code and it uses a harcoded 0.001 delta in the calculation 
>of the gradient and I didn't see that in any other 
>ImageToImageMetric.
>
>So my question is, does it actually work? Why is the delta hardcoded to 
>0.001?
>
>Thanks for your input,
>
>Damiaan
>
>  
>
>  
>