D.L. Schuler

Unsupervised classification using polarimetric decomposition and the complex Wishart classifier

The authors propose a new method for unsupervised classification of terrain types and man-made objects using polarimetric synthetic aperture radar (SAR) data. This technique is a combination of the unsupervised classification based on polarimetric target decomposition, S.R. Cloude et al. (1997), and the maximum likelihood classifier based on the complex Wishart distribution for the polarimetric covariance matrix, J.S. Lee et al. (1994). The authors use Cloude and Pottiers method to initially classify the polarimetric SAR image. The initial classification map defines training sets for classification based on the Wishart distribution. The classified results are then used to define training sets for the next iteration. Significant improvement has been observed in iteration. The iteration ends when the number of pixels switching classes becomes smaller than a predetermined number or when other criteria are met. The authors observed that the class centers in the entropy-alpha plane are shifted by each iteration. The final class centers in the entropy-alpha plane are useful for class identification by the scattering mechanism associated with each zone. The advantages of this method are the automated classification, and the interpretation of each class based on scattering mechanism. The effectiveness of this algorithm is demonstrated using a JPL/AIRSAR polarimetric SAR image.

On the estimation of radar polarization orientation shifts induced by terrain slopes

In recent studies, D. L. Schuler et al. (2000) applied polarimetric imaging radar-derived orientation angles to measure topography, and J. S. Lee et al. (2000) used orientation angles for polarimetric SAR data compensation, to ensure accurate estimation of geophysical parameters in rugged terrain areas. To support these applications, it is important to accurately estimate shifts in orientation angles induced by the azimuth slope variations. However, in many cases, inconsistency in the estimation of orientation angle shifts was encountered in several areas, introducing noisy and erroneous results. The present authors develop a unified analysis of estimation algorithms based on the circular polarization covariance matrix. The concept of reflection symmetry is used to explain the soundness of the circular polarization method and to show problems associated with other algorithms. L-band polarimetric synthetic aperture radar (SAR) images of Camp Roberts, CA, are used to substantiate this theory.

Measurement of topography using polarimetric SAR images

A processing technique for polarimetric synthetic aperture radar (SAR) data has been developed which produces profiles of terrain slopes and elevations in the azimuthal (or along-track) direction. This technique estimates the average shift in orientation angle of copolarization backscatter caused by azimuthal tilts of the scattering plane. Using P-band data, tests of this technique have been made for an area in the Black Forest near Villingen/Schwenningen in Baden-Wurttemberg, Germany. The radar measured slope and derived elevation profiles have low rms errors and high correlation values when compared with a stereo-photograph digital-elevation map (DEM) for the area. This algorithm is capable of adaptively making transitions from the forested areas to nearby regions with open-terrain. Subsequent tests of the algorithm have been conducted using polarimetric SAR L-band data for a mountainous, nonforested, region in the Mojave Desert (Ft. Irwin, CA) where an accurate DEM also was available. Complete elevation and slope mapping of the terrain in two dimensions using this technique is possible when azimuthal elevation profiles are produced throughout the range extent of the SAR image.

Scattering-model-based speckle filtering of polarimetric SAR data

A new concept in polarimetric synthetic aperture radar (POLSAR) speckle filtering that preserves the dominant scattering mechanism of each pixel is proposed in this paper. The basic principle is to select pixels of the same scattering characteristics to be included in the filtering process. To achieve this, the algorithm first applies the Freeman and Durden decomposition to separate pixels into three dominant scattering categories: surface, double bounce, and volume, and then unsupervised classification is applied. Speckle filtering is performed using the classification map as a mask. A single-look or multilook pixel centered in a 9 /spl times/ 9 window is filtered by including only pixels in the same and two neighboring classes from the same scattering category. This filter is effective in speckle reduction, while perfectly preserving strong point target signatures, and retains edges, linear, and curved features in the POLSAR data. The effect of speckle filtering on scattering characteristics, such as entropy, anisotropy, and alpha angle, will be discussed.

K-distribution for multi-look processed polarimetric SAR imagery

A K-distribution has been developed to characterize the statistical properties of multi-look processed polarimetric SAR data. The probability density function (PDF) was derived as the product of a gamma distributed random variable and the polarimetric covariance matrix. The latter characterizes the speckle and the former depicts the inhomogeneity (texture). For multi-look data incoherently averaged from correlated one-look samples, the authors found that, for better modeling, the number of looks has to assume a non-integer value. A procedure was developed to estimate the equivalent number of looks and the parameter of the K-distribution. Experimental results using NASA/JPL 4-look and 16-1ook polarimetric SAR data substantiated this multi-look K-distribution. The authors also found that the multi-look process reduced the inhomogeneity and made the K-distribution less significant.<<ETX>>

Surface roughness and slope measurements using polarimetric SAR data

In this paper, the circular polarization coherence, /spl rho//sub RRLL/, is investigated as a potential estimator of terrain surface roughness and small-scale slopes. The studies utilize microwave backscatter collected from 1) dielectric surfaces in an anechoic chamber and 2) a desert test site using P-, L-, and C-band NASA/JPL AIRSAR data. These experimental studies and supporting theory, indicate a sensitive decrease of |/spl rho//sub RRLL/| with increasing surface roughness ks over a range 0 /spl les/ ks /spl les/ 1. For the present studies this decrease is caused largely by the depolarizing effects of small-scale surface slopes in the azimuth direction rather than by volume, or multiple scatter. For cases when the scatter is reflection symmetric, the value of |/spl rho//sub RRLL/| depends on the surface roughness and on the local incidence angle. The dependence of |/spl rho//sub PRRLL/| on the local incidence angle is supported by theory and experimental results. For these same scattering cases, however, |/spl rho//sub PRRLL/| is independent of the surface dielectric constant. Estimation of the functional dependency of |/spl rho//sub PRRLL/| versus ks, for a mid-range incidence angle, has been carried out using roughness estimates derived from an empirical model.

Unsupervised classification using polarimetric decomposition and complex Wishart classifier

The authors propose a new method for unsupervised classification of terrain types and man-made objects using polarimetric SAR data. This technique is a combination of the unsupervised classification based on the polarimetric target decomposition (Cloude and Pottier, 1997) and the maximum likelihood classifier based on the complex Wishart distribution (Lee et al., 1994). The advantage of this approach is that clusters may be identified by the scattering mechanisms from the target decomposition. The effectiveness of this algorithm is demonstrated using JPL/AIRSAR and SIR-C polarimetric SAR images.

Target Detection and Texture Segmentation in Polarimetric SAR Images Using a Wavelet Frame: Theoretical Aspects

Theoretical aspects of a technique for target detection and texture segmentation in synthetic aperture radar (SAR) imagery using a wavelet frame are presented. Texture measures consist of multiscale local estimates of the following: 1) normalized second moment of the backscattered intensity and 2) variance of the wavelet-frame coefficients. This work is an extension of a method proposed in the image-processing literature. Novel issues, which are considered in the passage to radar imagery, are the influence of speckle on texture measures afforded by the wavelet frame and their dependence on polarization states (polarimetric texture). Regarding speckle, estimators that decouple the influence of speckle over texture are introduced and characterized by their expected value and variance. The response of the wavelet frame to discontinuities, which is an important issue in target detection problems, is addressed in terms of signal-to-speckle-noise ratio. The notion of polarimetric texture is revisited, providing a theoretical model that explains the dependences of texture measures on the polarization states. For one-point statistics, such model calls for a mixture of diverse polarimetric scattering mechanisms within the texture estimator support. For two-point statistics, the difference in spatial correlation properties among the polarimetric channels is called into play. To analyze these effects in polarimetric SAR data, a novel tool is introduced that is called the Wavelet Polarimetric Signature. The tool encapsulates, in graphical form, the dependence on scale and polarization state of the texture measure afforded by the wavelet frame. The theory exposed here underpins a method that has been proven successful and computationally attractive in a selected number of SAR thematic applications. It also sets the stage for the exploitation of novel target detection and textural segmentation capabilities based on polarimetric diversity.

Topographic mapping using polarimetric SAR data

A new remote sensing technique using polarimetric synthetic aperture radar (SAR) data has been developed which can measure terrain slopes in the azimuthal, or along-track, direction. Terrain elevation maps can then be generated by integrating these slopes. The processing of both single-pass, and orthogonal two-pass, datasets is investigated. When single-pass SAR data is used elevation groundtruth must be available for at least one point of each profile formed in the azimuthal direction. When orthogonal two-pass slope data is employed, the elevation surface may be generated as an iterative solution of the Poisson equation and only a single elevation tie-point is required. The results of two studies are presented. In the first study slopes have been measured using single-pass P-band data for a forested area in northern California. The elevation surface generated with this data has been compared with a Digital Elevation Model (DEM) produced by a C-band interferometric SAR. The second study uses orthogonal tw...

Polarimetric SAR image signatures of the ocean and Gulf Stream features

Polarimetric signatures and related polarimetric properties of microwave ocean backscatter are analyzed for both the ambient ocean and for ocean features such as those associated with the Gulf Stream. Interpretation of the polarimetric signatures for the ocean surface is accomplished using a tilted-Bragg theoretical model. This model is used to calculate the EM fields, to second order, which is necessary to compute the full Stokes matrix and, ultimately, the polarimetric signature. The polarimetric studies lead to a technique for potentially improving the visibility of all azimuthally traveling waves in real-aperture radar (RAR) images and very long waves in synthetic-aperture radar (SAR) images. This technique utilizes linear polarization signatures to maximize the instrument sensitivity to azimuthally traveling waves. Wave tilts create a modulation of the cell polarization orientation which, in turn, modulates the backscatter. Critical to the success of this technique is that the ocean polarimetric signatures be sharply peaked (i.e., returns be highly polarized). The polarimetric contribution to the overall modulation transfer function is evaluated. >