US Patent 9430499 Automated feature extraction from imagery

Embodiments of the invention are directed to a computer-implemented system and method of identifying human settlements in imagery comprising receiving an image, segmenting the image into a plurality of superpixels, analyzing statistical parameters of at least two or more of the plurality of superpixels, where the statistical parameters includes entropy data, and identifying groups of superpixels having at least a predetermined cluster density and a predetermined entropy. Some embodiments further include clipping the image to only include the identified groups of superpixels having the predetermined cluster density and entropy, analyzing statistical parameters of the clipped image, analyzing geometric factors of the clipped image, determining one or more settlements based on the statistical parameters and geometric factors of the superpixels, and identifying a shape and area of the one or more settlements based on the statistical parameters and geometric factors of the clipped image.