Uses KMeans clustering and the L*A*B* colorspace to extract "approximate human vision" dominant colors from an image. Optionally, use Euclidean Distance to map those dominant colors into preferred "color bins" for a search index facet-by-color solution.