Hello.
My goal is to generate a relevant segmentation that would delimit the main physiopnomic units (homogenous areas). Up to now, I have done several tests with qgis and orfeotoolbox 9 with the OTB Generic Region merging module : GenericRegionMerging - Orfeo ToolBox 7.0.0 documentation<https://www.orfeo-toolbox.org/CookBook-7.0/Applications/app_GenericRegionMerging.html> with those folowing settings
* input image : spot 6/7 satelite image (band blue, green, red, near infrared)
or Aerial french mapping agency images (50 cms bd ortho with green, red and near infrared).
* Homogenity criterion to use : Baatz & Schape
* Threshold for the criterion : 500
* Number of iterations`: 0
* Weight for the spectral homogeneity : 0.1
* Weight for the spatial homogeneity : 0.1
The result is rather interesting for certain areas, in particular crops areas, without too much vegetation. on the other hand, the method does not seems optimal, in particular for forest areas.
I'm taking the liberty to paste some screenshots :
the first SS showing interesting result on crop areas :
[cid:image001.png@01DB8D19.9DA95260]
A second SS showing the non relevant result on forest areas :
[cid:image002.png@01DB8D19.9DA95260]
Perhaps for these forest areas, the color/spectral criterion is not relevant. It would be better to generate / integrate texture information? As far as i understood, this GRM OTB module is based only on spectral/colour criterion but not take into consideration texture information.
Are there other qgis/grass/ orfeotoolbox or whatever other segmentation module that takes into account not only the color criterion (spectral) but also texture information?
If this is not the case, should i create some kind of new raster (add a band to my existing color infrared raster) dedicated to texture and/or ndvi information?
Thanks so much for your guidances.
(attachments)
Hello.
My goal is to generate a relevant segmentation that would delimit the main physiopnomic units (homogenous areas). Up to now, I have done several tests with qgis and orfeotoolbox 9 with the OTB Generic Region merging module : GenericRegionMerging - Orfeo ToolBox 7.0.0 documentation<https://www.orfeo-toolbox.org/CookBook-7.0/Applications/app_GenericRegionMerging.html> with those folowing settings
* input image : spot 6/7 satelite image (band blue, green, red, near infrared)
or Aerial french mapping agency images (50 cms bd ortho with green, red and near infrared).
* Homogenity criterion to use : Baatz & Schape
* Threshold for the criterion : 500
* Number of iterations`: 0
* Weight for the spectral homogeneity : 0.1
* Weight for the spatial homogeneity : 0.1
The result is rather interesting for certain areas, in particular crops areas, without too much vegetation. on the other hand, the method does not seems optimal, in particular for forest areas.
Perhaps for these forest areas, the color/spectral criterion is not relevant. It would be better to generate / integrate texture information? As far as i understood, this GRM OTB module is based only on spectral/colour criterion but not take into consideration texture information.
Are there other qgis/grass/ orfeotoolbox or whatever other segmentation module that takes into account not only the color criterion (spectral) but also texture information?
If this is not the case, should i create some kind of new raster (add a band to my existing color infrared raster) dedicated to texture and/or ndvi information?
Thanks so much for your guidances.
Hello Laurent,
You might want to check i.segment [0] within the core tools of GRASS and i.segment.uspo [1] for unsupervised parameter optimization within GRASS extensions. Also, there are nice papers by Grippa et al [2] and Georganos et al [3] that show the segmentation/OBIA workflow within GRASS.
To include texture rasters within segmentation, have a look at r.texture [4] tool in GRASS. What I usually do is to estimate textures from a panchromatic higher res band.
hth,
Vero
[0] https://grass.osgeo.org/grass-stable/manuals/i.segment.html
[1] https://grass.osgeo.org/grass-stable/manuals/addons/i.segment.uspo.html
[2] https://doi.org/10.3390/rs9040358
[3] https://doi.org/10.3390/rs10091440
[4] https://grass.osgeo.org/grass-stable/manuals/r.texture.html
···
Dra. Verónica Andreo
Investigadora Adjunta de CONICET
Instituto Gulich (CONAE - UNC)
Centro Espacial Teófilo Tabanera (CETT)
Falda del Cañete - Córdoba, Argentina
+54 3547 400000 int. 1153
https://veroandreo.gitlab.io/
Good afternoon,
Thanks so much for your usefull reply.
I’m taking the liberty to ask a little aditionnal question.
Still for segmentation needs, i plan to set for the input data, not only the multi spectral data (B,G,R,PIR) but also
ndvi, texture, DEM…And before to stack all these variables (and before to plan to run a PCA for reduce the information), could you confirm to that i need to rescale all the variables ? For example, for the NDVI (default Values from -1 to 1), i need to rescale its from 0 to 255 (if my multi spectral data is encoded on 8 bits unsigned). And i have to the same rescaling for others variables?
Thank so much.
···
De : Veronica Andreo veroandreo@gmail.com
Envoyé : mardi 4 mars 2025 23:33
À : Laurent CELATI lcelati@biotope.fr
Cc : grass-user@lists.osgeo.org
Objet : Re: [GRASS-user] Segmentation : how to add texture criterion ? (not only colour/spectral variable)
Hello Laurent,
You might want to check i.segment [0] within the core tools of GRASS and i.segment.uspo [1] for unsupervised parameter optimization within GRASS extensions. Also, there are nice papers by Grippa et al [2] and Georganos et al [3] that show the segmentation/OBIA workflow within GRASS.
To include texture rasters within segmentation, have a look at r.texture [4] tool in GRASS. What I usually do is to estimate textures from a panchromatic higher res band.
hth,
Vero
[0] https://grass.osgeo.org/grass-stable/manuals/i.segment.html
[1] https://grass.osgeo.org/grass-stable/manuals/addons/i.segment.uspo.html
[2] https://doi.org/10.3390/rs9040358
[3] https://doi.org/10.3390/rs10091440
[4] https://grass.osgeo.org/grass-stable/manuals/r.texture.html
El mar, 4 mar 2025 a las 12:20, Laurent CELATI via grass-user (<grass-user@lists.osgeo.org>) escribió:
|
|
|
|
|
|
Hello.
My goal is to generate a relevant segmentation that would delimit the main physiopnomic units (homogenous areas). Up to now, I have done several tests with qgis and orfeotoolbox 9 with the OTB Generic Region merging module : GenericRegionMerging — Orfeo ToolBox 7.0.0 documentation with those folowing settings
- input image : spot 6/7 satelite image (band blue, green, red, near infrared)
or Aerial french mapping agency images (50 cms bd ortho with green, red and near infrared).
- Homogenity criterion to use : Baatz & Schape
- Threshold for the criterion : 500
- Number of iterations`: 0
- Weight for the spectral homogeneity : 0.1
- Weight for the spatial homogeneity : 0.1
The result is rather interesting for certain areas, in particular crops areas, without too much vegetation. on the other hand, the method does not seems optimal, in particular for forest areas.
|
Perhaps for these forest areas, the color/spectral criterion is not relevant. It would be better to generate / integrate texture information? As far as i understood, this GRM OTB module is based only on spectral/colour criterion but not take into consideration texture information.
Are there other qgis/grass/ orfeotoolbox or whatever other segmentation module that takes into account not only the color criterion (spectral) but also texture information?
If this is not the case, should i create some kind of new raster (add a band to my existing color infrared raster) dedicated to texture and/or ndvi information?
Thanks so much for your guidances.
grass-user mailing list
grass-user@lists.osgeo.org
https://lists.osgeo.org/mailman/listinfo/grass-user
–
Dra. Verónica Andreo
Investigadora Adjunta de CONICET
Instituto Gulich (CONAE - UNC)
Centro Espacial Teófilo Tabanera (CETT)
Falda del Cañete - Córdoba, Argentina
+54 3547 400000 int. 1153
https://eu01.z.antigena.com/l/ZU13wcvQ5667aVP9D4sxjNbhD6LbbftuuvORk3hL1tmpYxTPazkeg7p8G7sk-5CKnXF9kYTehe~Nff-Aao4OI18GgOQYVTFxlORVAIcSAxYI5PHLGQvJof4iQL2HlI-Pr_tx6mvuLVnNWU6dzTfmI83H3Xkfer_-3N7VC44PmiVWmzKTfHs-DXUSN1x