07-21-2016 04:03 PM
I have an image and a shapefile with multiple polygons.
I would like to subset the image into each of the polygons.
How can I do this in Spatial Modeler?
I know I can do this with an AOI, but this requires manually converting the .shp to .aoi. I really need this to be automated...
Solved! Go to Solution.
07-24-2016 10:21 PM
This is close to what I am after, however I require the final result to be multiple individual images. Whereas this is outputting all of the subsets to one image.
07-24-2016 10:45 PM
The least computational approach may be to use an Iterator to yield a clip an image under each feature. You might have to use the CreateTempFile op within the Iterator.
If you're adventurous, you could push multicolor imagery through a RasterInput with type set to u1, and pipe that bitonal image through Clump. Following the customary Clump and Sieve recipe, either use a predefined set of Lookups to extract each clip or use an Iterator.
It's late here on the west coast and haven't tried those, but it is an interesting problem. Keep us posted if you find a solution.
08-03-2016 11:45 AM
01-07-2018 10:05 AM
Is there a modeler file in the old version *gmd? I can´t opend this *.
Noel is on the right track about starting with the Subset Image by Polygonal Geometries model and adding an Iterator.
Give the attached model a try.
01-08-2018 07:48 AM
Larry’s spatial model is only available as a *.gmdx file. The techniques used in his spatial model cannot be utilized in the old graphical Model Maker. I would recommend that you upgrade to ERDAS IMAGINE 2016 so that you can take advantage of the Spatial Modeler and everything that it has to offer.
Hexagon Geospatial Support
11-29-2018 05:46 AM
Hello Stephen and comunity!
I am facing a problem with the model shared by Larry. The model works pretty well..but it does not reduce the size of the image. I mean the chips that are created after the application of the model have the same dimension of the input file. The difference with the input is that the information is only available inside the chip while outside there are only 0. Even though this 0 are considered no data still the dimension of the raster is big because the input image was big. Do you have any idea on how to modify the model in order reduce the size of the chip in a way that will show only the real information ?