[Insight-users] Are these metrics based on "correlation ratio"?

[pinpress]

<from Luis: edit existing
itkCorrelationCoefficientHistogramImageToImageMetric.txx template file to
compute correlation ratio>


Hi Luis and all,

I am trying to modify existing ITK source code to implement correlation
ratio for image registration. I am testing the code now. However, the code
is very slow, and I suspect this is because of the following two reasons:

1. The code is not multi-threaded;
2. metric->SetNumberOfSpatialSamples( ) doesn't seem to have any effect

Am I correct, especially the 2nd one (since I know the first one was true)?

So my questions are:

1). How to multi-thread the class? Are there significant amount of work
involved to make this happen? I know Mattes MI and Mean Squares Differences
are multi-threaded, not others;

2. How come that the SetNumberOfSpatialSamples( ) function doesn't have any
effect on the computation? I would expect that the smaller number the
spatial samples, the faster it runs. What is the root cause for this then?

Thanks very much, for anyone who gives me new hint.
-- 
View this message in context: http://old.nabble.com/Are-these-metrics-based-on-%22correlation-ratio%22--tp28287905p28465520.html
Sent from the ITK - Users mailing list archive at Nabble.com.