AnsweredAssumed Answered

How to categorize a raster based on pixel count within a moving kernel?

Question asked by nil221292 on Feb 23, 2018
Latest reply on Feb 23, 2018 by Dan_Patterson

I have a classified raster with four land cover classes - built-up, vegetation, water and barren land.

Now, I want to categorize the built-up areas into three more classes - high built-up density, medium built-density and low built-up density.

I have reclassified my raster into two classes - built-up is 1 and vegetation, water and barren land has been merged into a single class 0.

For categorizing the built-up I am considering a moving circular neighborhood with a radius of 150 meters. My raster has a spatial resolution of 30 meters. As the kernel (neighborhood) moves, the central pixel is assigned a category every time.

And, the conditions for categorization are: high density if more than 50% of pixels within the neighborhood are built-up (class value = 2), medium density if 25 - 50% of pixels within the neighborhood are built-up (class value = 3) and low density if less than 25% pixels within the neighborhood are built-up (class value = 4).

I have been trying to use the model builder and the CON expression from the raster calculator. But have failed to get the desired output.

Please help if anybody has any idea as to how can this problem be solved!

Outcomes