Generating overlapping polygons from lines output from Table To Ellipse tool?

Question

I have a feature class of line-based ellipses generated by the Table to Ellipse tool that looks like this:
What I want is polygons of the individual ellipses, however tools like Feature to Polygon will create a polygon for every intersection.

My solution to this right now is a Search Cursor in Python that makes a layer from each line ellipse and then runs the Feature to Polygon tool on that layer. This is extremely slow and means I have a feature dataset with hundreds (soon to be thousands) of individual polygon feature classes in it instead of one file.

I saw from this question that there is a Python package, Shapely, that might run much faster than the arcpy tool in a loop. I can't believe, however, that there isn't some way to do this in ArcGIS. Can anyone think of a better solution?

Answer 1

Instead of using Feature to Polygon multiple times, you could use your SearchCursor to read the vertices of each ellipse and then create a polygon from the vertices.

shpin = "Feature class"
shpout = "output polygon FC"

#Get number of features in input.
entries = int(arcpy.GetCount_management(shpin).getOutput(0))
#Get name of OID field.
oid = [str(x.name) for x in arcpy.ListFields(shpin, "*", "OID")][0]

array = arcpy.Array()
polyvert = []

#Starting at 1 and running to entries +1, because Python ranges start at 0; 
#FC OID starts at 1. This can be changed to "for i in xrange(entries):" for a shapefile
for i in xrange(1,entries+1):
    #Use where_clause to loop over each feature in feature class.
    #explode_to_points is used to get coordinates of all vertices in ellipse.
    vertices = [row[0].firstPoint for row in arcpy.da.SearchCursor(shpin, 
                "SHAPE@", '"{0}"={1}'.format(oid, i), explode_to_points=True)]

    #Use list comprehension to add vertices individually to array.
    [array.add(x) for x in vertices] #Add point objects to array.

    #Create polygon from point array and append to list.
    polyvert.append(arcpy.Polygon(array)) 
    array.removeAll() #Remove data for next loop.

arcpy.CopyFeatures_management(polyvert, shpout) #Convert polygon array to shapefile
#Define same projection as input.    
arcpy.DefineProjection_management(shpout, arcpy.Describe(input).spatialReference)

If you don't need to maintain highly accurate output polygons, I recommend that you run generalize on your input before proceeding. I made a test shapefile of 25 ellipses with major,minor axes of 50,150. Each ellipse had over 21,000 vertices. Run time was 61 seconds. Using generalize set to 1 inch, the vertice count was reduced to 70 and run time was .71 seconds. The file size was also reduced from 7.8MB to 55KB.

If you are running this on thousands of ellipses that are expansive, run time can easily approach a few hours and you'll end up with a potentially very large output. Granted, if you are making ellipses in the first place, you probably want highly accurate (and vertex-dense) output polygons.