Chair: John D. Gorman, ERIM (USA)
Juan E. Garrido-Arenas, Centro Astronomico de Yebes
Jose M. Paez-Borrallo, ETSI Telecomunicacion-UPM
Alberto Barcia-Cancio, Centro Astronomico de Yebes (SPAIN)
Determination of the surface quality of large reflector antennas by direct methods like tape-theodolite has been revealed as not accurate enough for millimeter-wave operation. Indirect methods like holographic ones have been widely used. They are based on the Fourier Transform (FT) relationship between the far-field pattern and the field distribution in the aperture, whose phase can be used to obtain the map of axial deformations of the paraboloid by simple ray tracing. Measurement of the pattern phase requires a second antenna-receiver system and becomes difficult for high frequencies, so the possibility of recovering the aperture field from only- amplitude (or intensity) measurements of its FT (the pattern) has been studied and applied in radio telescopes measurements. We present a discrete model for the aperture that enables us to approach this problem from an array processing point of view.
Kie B. Eom, The George Washington University (USA)
In this paper, we consider the classification of radar signals by using stochastic models at different scales. The signal at a different scale is modeled by a hierarchical Autoregressive Moving Average (ARMA) model, and the features at coarse scales are extracted from the model without performing expensive filtering operation. The hierarchical modeling can increase the accuracy of radar signal classification by exploiting features at different scales. For radar signal classification, model parameters at five different scales obtained by hierarchical modeling are used as features. A minimum distance classifier is implemented, and is tested on real aperture radar signals.
Thomas L. Marzetta, Nichols Research Corporation (USA)
A polarimetric synthetic aperture radar (SAR) forms a complex vector-valued image where each pixel comprises the polarization-dependent reflectivity of a portion of a target or scene. The most common statistical model for this type of image is the zero-mean, circularly-symmetric,multivariate, complex Gaussian model. A logical generalization of this model is a circularly-symmetric, multivariate, complex Rician model which results from having a nonzero- mean complex target reflectivity. Direct maximum-likelihood estimation of the Rician model parameterrs is infeasible, since setting derivatives equal to zero results in an intractable system of coupled nonlinear equations. The contribution of this paper is a complete iterative solution to the Rician parameter estimation problem by means of the EM (expectation- maximization) algorithm.
Hung-Chih Chiang, The Ohio State University (USA)
Randolph L. Moses, The Ohio State University (USA)
Stanley C. Ahalt, The Ohio State University (USA)
This paper considers linear correlation filters used for image pattern recognition. First, we develop a statistical theory to predict the classification performance of a general class of correlation filters for wide sense stationary (WSS) clutter. This analysis includes as special cases the Synthetic Discriminant Function (SDF), the Minimum Variance SDF (MVSDF), and the Minimum Average Correlation Energy (MACE) filters. Second, we develop a modified filter design applicable to nonzero mean noise; this latter case occurs in many applications where the magnitude image is used for classification. We compare the performance of several filters on synthetic radar imagery.
X. Yu, SAIC
A.M. Chen, SAIC
I.S. Reed, Communications Science Institute (USA)
In [1] a matched-filter based detector was developed for the problem of detecting a 2-D target signal where prior information about the target pattern or template as well as the statistical properties of the clutter is limited. This was accomplished by an ad hoc substitution of the maximum likelihood estimate (MLE) of unknown clutter covariance matrix and the MLE's of the complex amplitudes of the significant features components of target into the matched filter test. This paper provides a new approach for the problem based on the generalized likelihood ratio (GLR) principle which maximizes the GLR function over unknown clutter covariance matrix and the unknown significant feature components of target signal to be detected. This new GLR test is compared with the matched-filter based test in [1] for performance. The feature mapping and representation which can be incorporated into the test to characterize the unknown target pattern are various, including the short time Fourier transform, the discrete cosine transform, and the discrete wavelet transform.