摘要:
The unique methodology and utility seeking patient protection in this application is using satellite imagery to quantify and derive harvest statistics of sun-drying crops.
High resolution imagery can be used to geospatially define known coffee drying basins however it is impractical for continuous observation as it is costly and infrequent.
The geospatially defined regions of interests of coffee drying basins can be matched with more frequent, lower resolution, multispectral satellite imagery (such as Sentinel-2). The signals can be tested against the known spectral signatures of washed and unwashed coffee to determine whether or not each pixel contains coffee.
The result will yield a classified region of interest which can be used to determine the quantity of drying coffee and the washed to unwashed ratio of a harvest. With regular monitoring across multiple temporal scenes the harvest's seasonality and historical change can be derived.
展开