Open mgiraldo opened 11 years ago
Have you looked at https://pypi.python.org/pypi/rdp/?
had not seen it. seems pretty new. will look into it
adding contents of #21:
I originally used R because I found code that reduces the amount of corners in a geographic polygon: http://rpubs.com/geospacedman/alphasimple
The function in the vectorizer is: https://github.com/NYPL/map-vectorizer/blob/9b9129b9d135c89b350700549794e7d0f2348af2/vectorize_map.py#L227
and the R code: https://github.com/NYPL/map-vectorizer/blob/9b9129b9d135c89b350700549794e7d0f2348af2/simplify_map.R
Ideally, we would eliminate R and have python code that produces the same output as R (one shapefile for each polygon that was simplified that is within the area threshold).
A shapefile-per-polygon is produced (thousands!) because it was the only way I could pass data from python-to-R-and-back but was also useful to find the average color of the polygon.
http://stackoverflow.com/questions/26303878/alpha-shapes-in-3d http://www.mrzv.org/software/dionysus/python/alphashapes.html http://www.mrzv.org/software/dionysus/python/overview.html#python-bindings http://doc.cgal.org/latest/Alpha_shapes_2/index.html http://stackoverflow.com/questions/17553035/draw-a-smooth-polygon-around-data-points-in-a-scatter-plot-in-matplotlib http://repositorium.sdum.uminho.pt/bitstream/1822/6429/1/ConcaveHull_ACM_MYS.pdf
Explore the possibility of making R unnecessary by implementing a Ramer–Douglas–Peucker or Visvalingam algorithm directly in Python. A Python implementation done by @migurski: https://github.com/migurski/bloch
See line 223: https://github.com/NYPL/map-vectorizer/blob/master/vectorize_map.py#L223 and
simplify_map.R
: https://github.com/NYPL/map-vectorizer/blob/master/simplify_map.Rvia @albertsun