Has anyone had the same issue? How do you handle the clouds in 
multitemporal raster segmentation?


Am Freitag, 15. Juli 2016 11:14:37 UTC+2 schrieb Caro So:
>
> Hi Jordi,
>
> what I expect as a result is that the boundary of the clouds is not 
> visible. For example I want to segment a multitemporal dataset (it does not 
> have to be a stack but can be single scenes) and in one image there are 
> clouds but in the other image there are no clouds. What I want to do is to 
> exclude the clouds from segmentation in the one image, but use the values 
> of the other image to 'fill' the cloudy areas with segments, as you can see 
> in the images attached.
>
> Is it possible to use a 'base segmentation' as additional input and to 
> simply refine the existing segments, i.e., to keep the boundaries of the 
> existing segments? Maybe this can be done by doing a segmentation on a 
> (nearly) cloud free scene, then use this segmentation as a base to segment 
> a second scene, a third scene and so on. The parameter "Update output 
> vector file, update existing layer" does not keep the existing segments, it 
> writes new ones.
>
> I hope you better understand my issue now?
>
> Kind regards
> Carolin
>
>
>
>
> Am Mittwoch, 13. Juli 2016 18:00:46 UTC+2 schrieb Jordi Inglada:
>>
>> Hi Carolin, 
>>
>> I understand that you are segmenting a 9 channel image which is composed 
>> from data from 3 different dates. I suppose that the clouds are not at the 
>> same position for all dates. 
>>
>> What do you expected as a result? What do you want to happen in the areas 
>> where clouds are present? 
>>
>> Jordi 
>>
>>
>> "'Caro So' via otb-users" <[email protected]> wrote: 
>> > 
>> > Hi there, 
>> > 
>> > I'm using the OTB Segmentation algorithm "Mean Shift Segmentation". My 
>> input raster is a Stack of 3 Sentinel-2 scenes, each having 3 bands 
>> (Principal component bands). The file format is GeoTIFF and the spatial 
>> resolution is 20m. 
>> > Everything runs fine except for the regions where there are clouds or 
>> masked out clouds. The values behind the region is NoData. However, the 
>> same problem occured when the NoData values were set to 0. See attached the 
>> JPEg images 
>> > which show the problem. 
>> > 
>> > I used the following parameters: 
>> > 
>> > Mean-Shift Segmentation 
>> > Spatial radius 10 
>> > Range radius 60 
>> > Mode convergence threshold 0,10000 
>> > Maximum number of iterations 20 
>> > Minimum region Size 20 
>> > 
>> > Tile-based large-scale segmentation with vector output 
>> > 
>> > Stich polygons ON 
>> > 
>> > How can one deal with clouds in the image so that the boundary is not 
>> segmented? 
>> > 
>> > Thanks for your help. 
>> > 
>> > Kind regards 
>> > Carolin 
>> > 
>> > 
>> > -- 
>>
>

-- 
-- 
Check the OTB FAQ at
http://www.orfeo-toolbox.org/FAQ.html

You received this message because you are subscribed to the Google
Groups "otb-users" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"otb-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to