CovAmCoh-Analysis: A method to improve the interpretation of highresolution repeat pass SAR images of urban areas

摘要

The main advantages of SAR (Synthetic Aperture Radar) are the availability of data under nearly all weather conditionsand its independence from natural illumination. Data can be gathered on demand and exploited to extract the neededinformation. However, due to the side looking imaging geometry, SAR images are difficult to interpret and there is aneed for support of human interpreters by image analysis algorithms. In this paper a method is described to improve andto simplify the interpretation of high resolution repeat pass SAR images. Modern spaceborne SAR sensors provide im-agery with high spatial resolution and the same imaging geometry in an equidistant time interval. These repeat pass or-bits are e. g. used for interferometric evaluation. The information contained in a repeat pass image pair is visualized bythe introduced method so that some basic features can be directly extracted from a color representation of three deducedfeatures. The CoV (Coefficient of Variation), the amplitude and the coherence are calculated and jointly evaluated. Thecombined evaluation of these features can be used to identify regions dominated by volume scatterers (e. g. leafed vege-tation), rough surfaces (e. g. grass, gravel) and smooth surfaces (e. g. streets, parking lots). Additionally the coherencebetween the two images includes information about changes between the acquisitions. The potential of the CovAmCoh-Analysis is demonstrated and discussed by the evaluation of a TerraSAR-X image pair of the Frankfurt airport. The me-thod shows a simple way to improve the intuitive interpretation by the human interpreter and it is used to improve theclassification of some basic urban features.