[Insight-developers] ChanAndVese Function: Scalart

[Luis Ibanez]

Hi Arnaud,

Thanks for the clarification.
It makes sense.

I agree with you that at this point,
adding the concept check is the best way to go.

I'll do that.


    Thanks


       Luis


---------------------
Arnaud Gelas wrote:
> Luis,
> 
> I guess it was following the LevelSetFunction implementation where all 
> computations are done with ScalarValueType (PixelType) and not with 
> PixelRealType (defined but not used).
> 
> I think it was assumed (at least in the RegionBased version) that the 
> input and output level set have float and double pixels.
> 
> Both of your suggestions sound good for me. But I think the easiest 
> solution is the second one, and in this period it may be preferable, right?
> 
> Arnaud
> 
> Luis Ibanez wrote:
> 
>> Arnaud, Kishore, Alex,
>>
>> The RegionBasedLevelSet function classes use the type
>>
>>
>>                 ScalarValueType
>>
>>
>> for returning most of the computation of the differential
>> equation terms.
>>
>>
>> This type is defined from the PixelType of the input level set.
>>
>> The type is used for performing floating point computations.
>>
>> There seems to be an assumption that the input (and output)
>> level sets must be images whose pixel types are float or
>> double.
>>
>> However, there is no ConceptCheck associated with this assumption.
>>
>>
>> Could you please verify if this observation is correct ?
>>
>> and if so,
>>
>>
>>    A) Should we introduce a RealScalarType in order to
>>       perform computations ?
>>
>>       (In which case, we will introduce the corresponding
>>        concept checks.)
>>
>>
>>    or
>>
>>
>>    B) Should we require that the input and output level set
>>       images use pixel types (float,double) ?
>>
>>
>>
>>    Thanks for any advice,
>>
>>
>>
>>       Luis
>>
>>
>>   
> 
> 
>