Polygonization of remote sensing classification maps by mesh approximation

The ultimate goal of land mapping from remote sensing image classification is to produce polygonal representations of Earth's objects, to be included in geographic information systems. This is most commonly performed by running a pixelwise image classifier and then polygonizing the connected components in the classification map. We here propose a novel polygonization algorithm, which uses a labeled triangular mesh to approximate the input classification maps. The mesh is optimized in terms of an l
₁
norm with respect to the classifiers's output. We use a rich set of optimization operators, which includes a vertex relocator, and add a topology preservation strategy. The method outperforms current approaches, yielding better accuracy with fewer vertices.