An Edge Aware Polarimetric Co-Variance Feature Matrix Generation of Geo-Coded and Quad-Polarized Single Look Complex Synthetic Aperture Radar Data

Abstract

Before the advent of microwave based imaging radars, most passive high resolution sensors were camera systems with detectors that were sensitive to either solar radiation or thermal radiation emitted from the earth’s surface. The Synthetic Aperture Radar (SAR) represented a fundamentally different technique for earth surface observations. A microwave based radar system is an active method of remote sensing that transmits a beam/out-burst of electromagnetic (EM) radiation which falls in the microwave region of the EM spectrum and this instrument is used to observe properties of the earth’s surface which were previously not detectable by ordinary photo-sensitive sensors (eg. optical, thermal). As an active system, SAR provides its own source of illuminating a target (microwave energy) and is not dependent on the light from the sun which most of the other type of sensors rely on, this permits a SAR based radar imaging for continuous day/night operation. Furthermore, neither clouds, fog, nor precipitation have a significant effect on microwave, thus permitting all-weather imaging capability. The net result is an instrument that is capable of continuously observing dynamic phenomena of ocean currents, weather patterns, changing patterns of vegetation, etc. However, all radar images appear with some degree of radar speckle i.e. graininess/salt and pepper texture in the image and is inherently present in any of the three modes (spotlight, scanSAR, stripmap) of acquisition. Speckle is a very serious and major issue in processing of SAR images and it is extremely difficult to go for machine interpretation and extraction of useful information from the mapped data. This problem of graininess in the image of an earth feature is caused by random constructive and destructive interference from the multiple scattering returns that occur within each resolution cell. SAR data has been used for a variety of applications (e.g. cartography, geologic structure mapping) for which qualitative analyses of the image products were sufficient to extract the desired information. However, to fully exploit the back-scattered information contained in a raw SAR data, quantitative analysis of the target back-scatter characteristics is required. Also, raw SAR data suffers from geometric distortion which arises from variation in the terrain elevation and pose a problem to side looking ranging instrument as in the case of a SAR system. In this paper, we have proposed a series of processing steps using which a very accurate polarized feature matrix values for a particular back-scatterer on earth’s surface can be obtained.