High Resolution Satellite Classification with Graph Cut Algorithms

摘要

In this paper, an unsupervised classification technique is proposed for high resolution satellite imagery. The approach uses graph cuts to improve the k-means algorithm, as graph cuts introduce spatial domain information of the image that is lacking in the k-means. High resolution satellite imagery, IKO-NOS, and SPOT-5 have been evaluated by the proposed method, showing that graph cuts improve k-means results, which in turn show coherent and continually spatial cluster regions that could be useful for cartographic classification.