Applied-GeoSolutions/lidar2dems

improve performance of gap-filling

matthewhanson opened this issue · 3 comments

Gap-filling, especially for large images, can take a long time. It uses numpy to fill the vales, and ends up interpolating for all points outside the boundary of data.

Two options, or perhaps both, to speed this up:
1 - implement gap-filling with a general GIPPY function
2 - incorporate mask so that interpolation does not occur outside site boundaries

related to issue #2

Gap-filling one product for Polygon 003 took 2 hours 50 minutes.

For DSM there is 1 product (max), and for DTM there are 3 (min, max, idw)

Gap-filling was improved with v0.2.0rc1 however, the options given here would still be viable to improve further.