=========================

Given input directories of before and after image tiles, calculates relative change by:

• Creating a grayscale image of the input bands
• Calculates edges (roberts cross method)
• Performs a principal component analysis on these two edge arrays
• The second principal component of the above is then cut into classes, and either plotted w/ side-by-side before and after images, or
• Vectorized into a geojson feature stream (useful for piping into tippecanoe for creation of Mapbox Vector Tiles)

git clone this repo and cd into it

then

pip install -e .

tiledelta.compareGreys(before_array, after_array, high_percentile, mid_percentile)

tiledelta comptiles [OPTIONS] FILEDIR COMPAREDIR

Options:
  -s, --sampling INTEGER (sampling stride to speed up vectorization)
  -f, --filetype TEXT (valid file extension)
  -p, --plotdir PATH (where to save plots to - optional)
  --help                  Show this message and exit.

NOTE: Input filenames need to fit this format!

{prefix}-{x}-{y}-{z}.{ext}

eg

tiledelta comptiles /Users/dnomadb/test_kath/ /Users/dnomadb/test_before/ -s 1 | tippecanoe -o test.mbtiles -z 15 -Z 11 -f

makes valid vector tiles expressing the degree of change for that location.