For us, this is relevant for land-use data, to check whether a certain area contains a certain vegetation type, or to get a list of the most common vegetation types in that area.
We will have to make a tradeoff between accuracy and performance. I don't think iterating over all pixels is necessary for most applications, we can just take a few samples. But both options should probably be provided.
For us, this is relevant for land-use data, to check whether a certain area contains a certain vegetation type, or to get a list of the most common vegetation types in that area.
We will have to make a tradeoff between accuracy and performance. I don't think iterating over all pixels is necessary for most applications, we can just take a few samples. But both options should probably be provided.
GDAL offers some of this (https://gdal.org/api/gdalrasterband_cpp.html#classGDALRasterBand), this might be worth looking into since it's probably well optimized (but probably also overkill).