Abstract:

A new algorithm of runways detection based on unsupervised classification is proposed. Firstly, initial sample templates are constructed from the original image with h/q decomposition. Then, the pixels in the original image are classified again with Bayesian classifier based on the initial sample templates. Thirdly,combining the property of polarization scattering and the weak backscattering feature of runways with Morphology filtering, suspected runway areas will be extracted from the above classification image. Using the runways structural features to identify suspected runway areas, the real runway area is detected finally. Multi-look fully polarimetric synthetic aperture radar (SAR) data acquired by U.S.UAVSAR systems is used to test the proposed algorithm. Experimental results show that the novel algorithm can detect runways effectively from complex scenes of the polarimetric SAR image and has a low false alarm rate and the detected results keep an intact structure and clear outlines.