deltares/geomorphometry.jl

Geospatial filtering algorithms

geomorphometry.jl's Issues

Is LocalFilters.jl needed?

Is there a reason to use LocalFilters.jl over ImageFiltering.jl with mapwindow ? Maybe there is a good reason that I'm not aware of.

Generalise types?

A lot of the methods defined here are limited to e.g. ::Matrix inputs.

It would be nice if we could pass through different array types using ::AbstractMatrix, and use similar to define new arrays so the wrapper type is maintained.

TagBot trigger issue

This issue is used to trigger TagBot; feel free to unsubscribe.

If you haven't already, you should update your TagBot.yml to include issue comment triggers.
Please see this post on Discourse for instructions and more details.

If you'd like for me to do this for you, comment TagBot fix on this issue.
I'll open a PR within a few hours, please be patient!

Rename to `Geomorphometry.jl`

Also in terms of package name, GeoMorphology?

Originally posted by @evetion in #22 (comment)

Add support for direct use of geographic data

I'd like to calculate slope and curvature from a geographic DEM (raster dimensions are in degrees). I'd like to do this without retrojecting the DEM. I think the only way to do this is to calculate component values (slope/curvature along lat/lon) then to scale each component by the local meters per degree lat/lon. Given this, would it make sense to add an option to relevant functions to return component values?

Implement SPMF

An improved simple morphological filter for the terrain classification of airborne LIDAR data
https://doi.org/10.1016/j.isprsjprs.2012.12.002

Use less acronyms

Unless you're really deep in the field MDG, TPI, smf, pmf, psf and apsf are pretty obscure, an not really googleable.

It could be nice to having these written out a little more so people know whats here. Even just the file names would make it more transparent what is in this package.

Should Geomorphometry.jl also include the ability for sparse calculations?

I have a global high-resolution DEM (cop30) and a ton of profile data (ICESat-2) and I'd like to calculate the slope and curvature at each ICESat-2 measurement point. I could do this by calculating slope and curvature everywhere but it would be much faster (less allocations) to simply calculate slope and curvature at each point. This is a utility that I often need but have not found.

Looking into Stencils.jl it seems like this shouldn't be too difficult... the steps would look like this

1. lazily read in profile data (probably Rasters.jl)
2. eagerly read data cropped to extents of profile data + a buffer for calculation of slope, curvature, ect
3. cast copped data to a SencilArray = A
4. loop though each intersecting index of A and apply functions

I'm guessing that this is functionality that others have built or would want but it's not clear where such functionality should live. Given that Geomorphometry.jl is home to most of the funtions I'm wondering if it should live here.

Thoughts?

Add support for slope along dimension (Sobel operator?)

For a lot of my work I need slope and other parameters (like curvature) along individual dimensions (i.e. slope in x and slope in y). For slope I think this is identical to the Sobel operator (https://en.wikipedia.org/wiki/Sobel_operator).

Interface for filter implementations

I wasn't sure if this is appropriate here or a different package, but wanted to start somewhere.

Sometimes when I am filtering a point cloud I don't want the surface yet, but want to identify/update the classification for what points are ground. I think it would be great to have some kind of generic interface like GeoStats/DiffEq/others where I can swap out the algorithm easily.

classify(pc::PointCloud, alg::GroundFilter) or
classify(pc::PointCloud, alg::VegetationFilter)

and then if I don't care about the points and want to go straight to DEM it classifies the point and returns the surface.

filter(pc:PointCloud, alg::GroundFilter)

Does this make sense?

I know there is an update!() functionality in LasIO that will update bounding box. Maybe the classification code lives in LasIO/LazIO and the filtering code lives here?

pssm broke in latest release

As the slope in cb5ca1b#diff-f272a4f29c1eec1436fa85b81604ffe52d4c9985555e8b3259cee62ddab28be1R18 needs to be scaled within 0-1.

Alternative and related packages

• https://github.com/JuliaImages/ImageMorphology.jl (general image morphology)
• https://github.com/emmt/LocalFilters.jl (fast mapwindowed operations)
• https://github.com/rafaqz/Stencils.jl (custom mapwindowed kernels)
• https://github.com/hellemo/FastMarching.jl (for spread, cost distance)