concave hulls using shapely and scipy

concave_hulls.ipynb

You can still try, but you may not be able to find an acceptable alpha value. It's been a few years since I was working on this but if I recall correctly, the delaunay triangulation doesn't generate enough triangles to be later pruned, or at least the space between them is not great enough.

# Area of triangle by Heron's formula
area = math.sqrt(s*(s-a)*(s-b)*(s-c))
circum_r = a*b*c/(4.0*area)
if circum_r < 1.0/alpha:
    # Keep these edges

Those areas need to be low enough to be less than the inverse of the alpha parameter. I'd step through this in a debugger with your shapefile and see what kind of values you're getting for each triangle to see if you can find a better set of parameters.

FYI for anyone running this notebook today (unless Descartes receives an update), you need to patch Descartes as described in this stackoverflow answer or the plot_polygon won't work https://stackoverflow.com/questions/75287534/indexerror-descartes-polygonpatch-wtih-shapely

I get the error "AttributeError: 'Delaunay' object has no attribute 'vertices'"

Is the .vertices method deprecated for Delaunay?

Yeah looks like it scipy/scipy@99cc995

Switch to tri.simplices

Yeah looks like it scipy/scipy@99cc995

Switch to tri.simplices

That fixed it. Thanks!

I am trying to get this to work for two offset sine waves. However, the best I can get is the second image. After that, increasing the alpha parameter creates a segmented geometry. Perhaps my ordering of points isn't in the proper format for this function?