[Insight-developers] Hisotgramming problems with large images

[Richard Beare]

Hi,
I'm playing with some largish images (1395 x 1130 x 562) which contain
small floating point values (0->6e-7). I'm experimenting with
histogramming methods (otsu thresholding + creating my own histogram
based on ImageHistogram2 in the examples), but get the following
error:

itk::ERROR: DenseFrequencyContainer(0x514a0a0): Frequency container
saturated for Instance

Which seems to result from the number of entries in a bin being larger
than the largest integer a floating point value is able to represent.
Is there any easy way around this?

I was hoping to avoid messing around with scaling and subsampling at
these early stages.

Thanks