06-28, 13:30–14:00 (Europe/Tirane), UBT D / N113 - Second Floor
We will discuss the algorithms inside geowarp, a high-performance and very low-level JavaScript library for reprojection, resampling and cropping of data from GeoTIFFs and other rasters. This talk will be at the abstract algorithmic level and is suitable for everyone. Here are some of the various algorithms that we will discuss:
- proj-turbo: fit an unknown reprojection function to a simple affine transformation
- fast-min/fast-max: calculating the range of your raster data leveraging the theoretical limits of the data types
- near-vectorize: automatically determining the optimal resampling algorithm based on relative pixel size
- dufour-peyton-intersection: calculate the pixels of an arbitrary raster inside an arbitrary polygon
- various resampling techniques including nearest, bilinear, vectorization, and box-based statistical methods
Daniel Dufour is the CEO of GeoSurge, a geospatial tech company with a focus on data compression, natural language processing, remote sensing, and visualization. He created or co-created several open-source geospatial projects, including geotiff.io, georaster, georaster-layer-for-leaflet, geoblaze, geowarp, proj4-fully-loaded, and proj-turbo.