I am working on a project where I take a large raster file (1.5GB+), put it through a number of image-processing operations, and finally convert it to vector data. Being that the raster is large, I am first breaking the raster up into smaller subsets (ie tiles) to aide in running the operations in parallel distributed amongst numerous machines. At the end of the chain of operations, each process/machine runs gdal_polygonize on the raster data to create and insert vector data into a shared PostgreSQL/PostGIS database. Rather than inserting each tile's polygons as they are, I'd prefer to first query if they can be merged with any tiles that either overlap (in the event that create subsets with overlap) or have coinciding edges and either update a row's geometry upon overlap/toucing or insert a new row if no data overlaps/touches. To complicate the matter, polygons have a depth value to be used as a criteria to determine if polygons should be merged.
Can anyone provide insight on a sane workflow to achieve this?
أكثر...
Can anyone provide insight on a sane workflow to achieve this?
أكثر...