Showing results for 
Search instead for 
Do you mean 

Clump and Sieve 2016

by Technical Evangelist on ‎07-01-2016 11:37 AM - edited on ‎02-21-2020 07:18 AM by Community Manager (3,653 Views)

Download model



This Model performs a traditional Clump and Sieve operation.  Clump is a commonly used technique for converting thematic class values in a raster into uniquely numbered "polygons" representing contiguous groups of the original class values. The Sieve component then removes clumps which are smaller than a user-specified size tolerance (measured in pixels, but this could easily be specified in ground units and converted to input-equivalent pixels in a Spatial Model). Additionally the Model extracts the color scheme used by the original input Thematic image classes and assigns it to the output sieved clumps, as well as transferring an attribute field containing the original class DN number (Original_Value) and the original class names (Class_Names).


Input Landcover Classification on the left, result of sieving clumps smaller than 7 pixels on the right (with background View color set to bright green)


The Model also provides a good example of how to use the :: Read operator and Raster Attribute Output operator.


Due to a couple of issues in older releases this Spatial Model requires ERDAS IMAGINE 2016 or later to run.


Note that the model assumes that an attribute field called Class_Names exists in the input thematic image and will attempt to transfer that attribute field to the output clumped and sieved image. An attribute filed of this name is commonly created by ERDAS IMAGINE when performing image classification operations. If no attribute field of this name exists in the input image that leg of the model will fail, but the overall process will still complete. If class names exist in a differently named attribute field alter the port on the :: (Read Classes) operator appropriately. Similarly, if you wish to transfer other attribute fields, simply copy and paste extra :: / Lookup / Raster Attribute Output rows and set them up in a similar fashion to those shown.





Input parameters: 


Thematic In: Name of input Thematic image to be Clumped.

Filename Out: Name of the output sieved thematic image.

Threshold: Minimum clump area (in pixels) below which to set the clump class to background. I.e. any clump containing less than this number of pixels will be removed from the thematic image.





by Technical Evangelist
on ‎03-01-2019 09:48 AM - last edited on ‎02-21-2020 08:57 AM by Community Manager

I had the need to build a Spatial Model which performed a Clump and Sieve as above, but that altered the output DN values to be those of the original input image rather than the Clump numbers. So I built the following and will try to attach the model as well. The Model also shows a different way to re-attach all attribute fields from the original input image (which is possible because of mapping back to original DN value):



Download Model