In this paper we focus on image-contrast optimization between two rough-surface classes. Our approach is based strictly on polarimetric filtering, and therefore, no digital image-processing techniques are employed. The approach is tested on a complete polarimetric synthetic aperture radar (POL-SAR) image of the San Francisco Bay area. The data have been taken with the National Aeronautics and Space Administration – Jet Propulsion Laboratory CV-990L-band POL-SAR system, where eight real numbers (complex elements of a 2 × 2 polarization scattering matrix) are associated with each image pixel. Optimal transmitted polarizations (corresponding to maxima or minima of reflected energy) are found for each image pixel, and the results are analyzed statistically via a set of joint two-dimensional histograms. This is done for both of the rough-surface classes. The image response to the "optimal" incident polarization is then simulated digitally by adjusting the receiver polarization according to the modes of the histograms. The corresponding images are computed and displayed with significant image-contrast improvement.
We focus on image contrast optimization between two rough surface classes, which is based strictly on polarimetric filtering and, therefore, no digital image processing techniques are employed.The approach is tested on a complete polarimetric synthetic aperture radar image of the San Francisco Bay area (NASA/JPL CV-990 Gband POGSAR data). Optimal transmitted polarizations are found for each image pixel and the results are analyzed statistically via a set of joint 2D histograms. This is done for both of the rough surface classes. The image response to the "optimal" incident polarization is then simulated digitally by adjusting the receiver polarization according to the modes of the histograms.The corresponding images are computed and displayed with significant image contrast improvement. I. 1NTRoDUcT1mThis paper addresses the problem of coherent image contrast optimization between two rough surface classes.we focus on such contrast which is due to the differences in polarimetric scattering from one rough surface to another. The novelty of this problem is due to the combination of coherent imaging and polarimetric scattering. The former introduces speckle reduction as a major issue while the latter provides the full scattering matrix (i.e., complete polarization information) per image pixel. The second and equally important task of this work is to develop efficient statistical tools for polarimetric image data analysis and speckle reduction techniques.Speckle has long been recognized as the main problem of coherent imaging (1) and many processing techniques have been advanced to overcome it.The vast majority of these techniques, however, are of a scalar nature sinply because vector/laatrix imaging data are 80 sparse and have become available only very recently. Such data, taken with the W J P L CV-990 dual-polarization L-band (1.225 GHZ) SAR (Synthetic Aperture Radar) system, have been made available to us.Here, we investigate the potential of a strictly polarimetric image filtering which takes full advantage of the matrix data provided on a pixel by pixel basis, and complements the existing scalar contrast optimization and speckle reduction techniques.We wish to stress from the outset that our goal is contrast optimization (with the corresponding speckle reduction) withut the help of incoherent averaging over pixels or "looks", because of the corresponding loss of spatial or temporal resolution.At first glance, speckle reduction is impossible without incoherent averaging but further consideration shows that it is so only for scalar data. Indeed, taking "projections" onto the receiver direction in the polarization space decreases amplitude fluctuations and an image appears less speckled. The goal of this paper is to find such a choice of the polarization projection which makes a given rough surface class least speckled and, by doing so, to improve the image contrast between two given classes.The paper is structured as follaws: a brief description of basic polarimetric definitions is provided in Section 11, while in...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.