A collection of routines for compositing Sentinel-2 images.

Algorithm for compositing Sentinel-2 images based on the S2GM Algorithm. The algorithm is currently set up for processing images processed with STORM.

• Each pixel is an actual observed pixel selected from all possible observations (not mean, etc.)
• Custom time interval (best results for monthly or seasonal time span)
• Set geographical extents at input
• Spatial resolution: 10m or 20m
• Using two different methods for compositing:
  • Short Term Composit (STC) for <4 valid pixels
  • Medoid for >= 4 valid pixels

S2GM website: https://s2gm.sentinel-hub.com/

Parallelised version of composite_s2gm.py (using concurent.futures module for processing on multiple cores).

Initial version of the compositing algorithm. Processing on single core, very slow. (REMOVE WHEN FINISHED)

Early version of the algorithm, based on the mosaicking algorithm for PROBA-V written by Klemen Cotar. Much faster then S2GM but doesn't select an actual pixel, i.e. each band is processed separately from each other.

A quick routine to obtain RGB preview of a GeTIFF.

Used for development. (REMOVE WHEN FINISHED)

Used for development. (REMOVE WHEN FINISHED)

Technical report: Testing of the compositing algorithm for Sentinel-2 products