[ITK-users] FastMarching UpwindGradient Filters difference

[Bradley Lowekamp]

Arnaud,

Thanks for the info on this.

I am looked through the code in the FastMarchingImageFilterBase [1], it looks like its using std::vector a surprising number of times, and that it is getting re-allocated very frequently. Some cases look like they should be stack allocated templated array. Do you know why this class is this way? This may be the reason why it's so slow in debug mode, i.e. too much dynamic allocation on a per-pixel basis.

Brad


[1] https://github.com/InsightSoftwareConsortium/ITK/blob/master/Modules/Filtering/FastMarching/include/itkFastMarchingImageFilterBase.hxx


On Jan 21, 2015, at 9:33 AM, Arnaud Gelas <arnaudgelas at gmail.com> wrote:

> Hi Brad,
> 
> FastMarchingUpwindGradientImageFilter is from ITKv3 while the other one is from ITKv4. Note that in terms of features, FastMarchingUpwiindGradientImageFilter would be a subset of the “Base" one…
> 
> Arnaud
> 
> 
> 
>> On 21 Jan 2015, at 15:09, Bradley Lowekamp <blowekamp at mail.nih.gov> wrote:
>> 
>> Hello,
>> 
>> I am trying to figure out the difference between the FastMarchingUpwindGradientImageFilterBase[1] and the FastMarchingUpwindGradientImageFilter[2].
>> 
>> The first thing to notice is that the "Base" class is NOT a parent of the other. Secondly, the non-Base implements a stopping criteria of reaching  a number of target nodes. This is now incorporated as an optional stopping criteria [3], in the framework. Then both of the filters add the upwind gradient image as an output, with what appears to be duplicated code. 
>> 
>> I don't see a real difference here. I think one should be deprecated.
>> 
>> Brad
>> 
>> [1] http://www.itk.org/Doxygen/html/classitk_1_1FastMarchingUpwindGradientImageFilterBase.html
>> [2] http://www.itk.org/Doxygen/html/classitk_1_1FastMarchingUpwindGradientImageFilter.html
>> [3] http://www.itk.org/Doxygen/html/classitk_1_1FastMarchingReachedTargetNodesStoppingCriterion.html
>