I generated peaks of tree crowns as an individual pixels from DSM dataset, but there are some trees that have more than 1 peaks. I would like to integrate them based on the threshold value, for example I want to integrate 2 or 3 neighboring pixels that lay in 1 square meters and create one in the middle. I know Integrate tool can do this for vector dataset, but if I convert peaks to vector and re-convert to raster, there is a change in raster dataset which affects further analysis. My question is is there any tool in ArcGIS similar to Integrate which works for raster dataset. Or is there any tool to integrate neighboring pixels in any GIS software. Thanks in advance for any suggestions!
In ArcGIS you can use majority or minority filters with grid (4*4 or 8*8.... etc), for Shapefiles you can use the merge, clip, union, buffer or reshape functions (1 meter, 2 3 ....... etc) as mentioned attached.
https://www.wou.edu/las/physci/taylor/es341/geoprocessing_ArcGIS.pdf
https://blogs.esri.com/esri/arcgis/2010/08/12/merging-multiple-shapefiles-into-a-single-shapefile/
http://gis.stackexchange.com/questions/133975/reshaping-polygon-shapefile-using-arcgis-for-desktop
In ArcGIS you can use majority or minority filters with grid (4*4 or 8*8.... etc), for Shapefiles you can use the merge, clip, union, buffer or reshape functions (1 meter, 2 3 ....... etc) as mentioned attached.
https://www.wou.edu/las/physci/taylor/es341/geoprocessing_ArcGIS.pdf
https://blogs.esri.com/esri/arcgis/2010/08/12/merging-multiple-shapefiles-into-a-single-shapefile/
http://gis.stackexchange.com/questions/133975/reshaping-polygon-shapefile-using-arcgis-for-desktop
Hi,
as mohamed has mentioned, majoritz filter would help zou to aggregate neighbouring pixels to the value which more often occurs within the kernel (4X4 or 8X8).
Alternatively, I suggest you another approach within ArcMap.
As your crowns are probably roughly circular "aggregation" of pixels, try to use Focal Statistics tool (Spatial Analysis toolbox). If I understood, you applied Raster Calculator (or Reclassify) to set a threshold and extract crowns. So, your raster should have pixels with value=1 (crown) and value=0 (non-crown).
Focal Statistics is a neighbourhood operator and it allows you more flexibility on kernel size, shape, function to apply and you can also use a weighted custom kernel !
Thus, try to use a circular kernel with the function "majority" or (better) write a kernel ASCII text file yourself with a small spatial extent (number of horizontal cells * cell size).
Give it a try!
Regards,
Nic
First thing that comes to mind is GRASS GIS command r.neighbors
https://grass.osgeo.org/grass70/manuals/r.neighbors.html
that offers a variety of way to "integrate" neighboring pixels.
See also r.mfilter
https://grass.osgeo.org/grass70/manuals/r.mfilter.html
Another tip here
http://gis.stackexchange.com/questions/1095/how-to-apply-a-majority-filter-in-grass-gis
Hope this helps,
Luigi
- Badges
- Science topic
- Similar topics
- Geoscience
- Geoinformatics
- Geographic Information System