/HabitatSampler

A procedure for autonomous reference data sampling, model training and image classification.

Primary LanguageHTMLGNU General Public License v3.0GPL-3.0

GitDocs/Logo.png

Procedure on Autonomous Sampling and Reductive Learning in Imagery

How to use

1. Stepwise Procedure

  • You need R to run the master script: HabitatSampler_v02.r
  • Within the master script a step by step procedure is executed: HabitatSampler_Usage.txt
  • This is the routine: HabitatSampler_v2.0

2. R package

  • You need R to install the package HaSa that includes all functions and test data
  • devtools::install_github("carstennh/HabitatSampler", subdir="R-package", build_vignettes = TRUE)
  • Sometimes there are problems, then do 1. devtools:: install_version("velox", version = "0.2.0", repos = "https://cran.uni-muenster.de/")
  • For Windows operating systems the Rtools are needed
  • library(HaSa) and list datasets: data(package="HaSa") and functions: lsf.str("package:HaSa") or use library(help="HaSa")
  • there are information available about programm execution and function behavior in Rmarkdown: HabitatSampler_Usage

Input

  • Image File as Raster Layer Stack (e.g. Satellite Time Series, RGB Drone, Orthophoto)
  • Reference File (e.g. spectral-temporal profiles or point shape; one profile or point per category)
  • Class Names (the categories that are defined to be delineated in imagery)

Output

  • Interactive Maps of habitat type probailities

GitDocs/figure_1.png

  • Classified Image of chosen categories
  • Sample Distribution of sampled categories
  • Spatial Statistics of categories distribution
  • the categories are refferred to as habitat types

GitDocs/figure_2.png

Key Features

  • the algorithm provides a set of reference samples for each habitat type
  • the algorithm provides an ensemble of calibrated machine learning classifiers for each habitat type
  • the algorithm provides a map of habitat type probabilities
  • the algorithm is optimzed for broad-scale satellite image time series (pixel size > 10m)
  • the alogrthm can be applied on variable image categories in complex scenes
  • the algorithm is tranferable to variable input imagery

Citation

Neumann, C. (2020): Habitat sampler—A sampling algorithm for habitat type delineation in remote sensing imagery. - Diversity and Distributions, 26 (12), 1752-1766. https://doi.org/10.1111/ddi.13165.

Credits

HaSa was developed by Carsten Neumann (Helmholtz Centre Potsdam GFZ German Research Centre for Geosciences) within the context of the NaTec - KRH project funded by the German Federal Ministry of Education and Research (BMBF) (grant number: 01 LC 1602A).

The test data represent pre-processed Copernicus Sentinel-2 satellite imagery (ESA 2018). Pre-processing was done using GTS2 and AROSICS.