topic Re: Image Resampling and Edge Effects in Spatial Statistics Questions

Image Resampling and Edge Effects

JohnWeidler — Mon, 24 Sep 2018 17:09:48 GMT

Hello,

I want to resample an image, using something like neighborhood statistics, but I know this will lead to edge effects. Is there a tool in ArcMap or Pro that will resample an image while adjusting for edge effects? Thanks.

Re: Image Resampling and Edge Effects

DanPatterson_Retired — Mon, 24 Sep 2018 17:20:40 GMT

What type of edge affects do you want to accommodate for?

This means that you have no adjacent data that you can use?

And/or you don't want to clip the outputs to remove the edge areas?

The type of treatment used is largely not specified.

The simplest way is to pad the perimeter with nodata values equivalent to 1/2-ish the analysis window size

Re: Image Resampling and Edge Effects

JohnWeidler — Mon, 24 Sep 2018 18:20:24 GMT

There's no adjacent data to use, and I'd rather not clip out edge areas if I don't have to. How do you pad the perimeter with specified values, is there an edge extension tool?

Re: Image Resampling and Edge Effects

DanPatterson_Retired — Mon, 24 Sep 2018 20:04:33 GMT

there is no edge tool.

You set the extent of the raster slightly larger than the raster. Assuming that the raster is rectangular and axis-oriented, find the current extent, then set the Left, right, top and bottom appropriately so you have a border of nodata values.

Numpy has specify tools for 'padding' arrays (aka, rasters) and functions that perform moving/rolling functions that account for nodata.

I suspect most of esri's functions account for edge effects behind the scenes, so before spending a lot of time on this, compare a simple neighborhood mean with a 3x3 moving window with and without the nodata collar.

Re: Image Resampling and Edge Effects

JohnWeidler — Mon, 24 Sep 2018 20:40:46 GMT

Thanks!