[GRASS-dev] mask for classification

[Quentin Page]

Moritz Lennert a écrit :
> On 11/05/09 09:51, Quentin Page wrote:
>> hi
>>
>> I try to create a mask for the pixels which have been misfiled. For 
>> this I made a mapcal with the pixels of the original image are equal 
>> 0, but when I have relaunched a classification, the result is 2 
>> classes (pixel to zero and others).
>
> I personally do not really understand the question (but maybe I'm just 
> dense), so maybe you should reformulate it more clearly. Start by 
> giving us your entire workflow with the commands you use and the 
> (unexpected) result.
>
> Moritz
hi Moritz
thank you for your answers

my workflow are

r.mapcalc "mask246clas6red at ventoux_sud_quentin=if((classif6groupIN246 at ventoux_sud_quentin==4||classif6groupIN246 at ventoux_sud_quentin==5),red246_96 at ventoux_sud_quentin,0)"
r.mapcalc "mask246clas6green at ventoux_sud_quentin=if((classif6groupIN246 at ventoux_sud_quentin==4||classif6groupIN246 at ventoux_sud_quentin==5),green246_96 at ventoux_sud_quentin,0)" 
r.mapcalc "mask246clas6blue at ventoux_sud_quentin=if((classif6groupIN246 at ventoux_sud_quentin==4||classif6groupIN246 at ventoux_sud_quentin==5),bleue246_96 at ventoux_sud_quentin,0)"

  after

i.group input=mask246clas6red at ventoux_sud_quentin,mask246autoclas6green at ventoux_sud_quentin,mask246autoclas6bleue at ventoux_sud_quentin group=246groupstep2 subgroup=246subgroupstep2

i.cluster group=246groupstep2 subgroup=246subgroupstep2 classes=4 sigfile=sigClassifstep2classe3

i.maxlik group=246groupstep2 subgroup=246subgroupstep2 sigfile=sigClassifstep2classe3 Classifgrou246etape2IN reject=Classifgroup246etape2REJECT

the resulte is 2 classes 
1 class the pixel which was classes 4 or 5 in the last classification
2 the others

I found a method for classifed the pixels which were classified incorrectly in the last version.

thank a lot