[Insight-users] Re: some requested changes

[rc3@doc.ic.ac.uk]

Hi Samson,

The primary reason why the upper bound needs to be increased is that the 
itk::Histogram class cannot hold samples equal to the upper bound value set for 
the histogram. This is how the itk::Histogram class is designed (as was qutoed 
in an earlier message). As the itk::Histogram class can hold samples equal to 
the lower bound, there is no corresponding need to decrease the lower bound.

The histogram metrics don't take into errors introduced by rounding.

Thanks,

Raghavendra

Quoting Samson Timoner <samson at bwh . harvard . edu>:

> 
> Raghavendra,
> 
>  It is unclear to me why there needs to be an "UpperBoundIncreaseFactor"
>  but not a "LowerBoundDecreaseFactor".
> 
>  I would expect if there can be round off errors which can accidentally
>  make the numbers to a histogram too high, there can also be round off
>  errors that can accidentally make the numbers to the histogram too low.
> 
>  Is this logic correct?
> 
> -- Samson
> 
>