Principles of Modern Radar - Volume 2 1891121537

13.4 Radar Applications of Polarimetry 611synthetic aperture radar (SAR) polarimetric imaging data. Several authors have enhancedHuynen’s decomposition and demonstrated effective separation of target return from clutter[2, 5, 11, 18, 38, 40, 41]. Barnes [42] and Holm [43] developed a target decompositionapproach in which the covariance or density matrix representing a partially polarized returnfrom the time-varying target is formulated as a sum of three components: (a) a mean targetcomponent (matrix) for which the return is completely polarized; (b) a component representinga totally unpolarized return; and (c) a component representing a partially polarizedreturn. The Barnes and Holm decomposition has been shown to yield clutter discriminationcomparable in accuracy to that of the Huynen approach [11]. Cloude and Pottier [40]discussed a decomposition in which a 4 × 4 averaged coherency matrix for noise targethas been diagonalized by calculating its eiegenvalues and eigenvectors. The diagonal formof the coherency matrix has been shown to result in statistically independent target vectorswherein the effect of noise (or target fluctuations) is removed. The reader is referred toexamples in [11] for a discussion of the clutter discrimination potential of this approach.13.4 RADAR APPLICATIONS OF POLARIMETRYTarget detection, discrimination, and recognition pertain to the process by which the radarseparates and classifies the return from a resolution cell as the response due to target(object of interest), clutter (objects of lesser interest), or noise (see Chapter 9 in [3]). Inthe detection process, well-known techniques for thresholding noise and clutter amplitudeor exploitation of Doppler frequency are used to effect an initial separation of target andclutter returns [4]. After detecting a target, the radar attempts to discriminate betweenpotential targets of interest and strong target-like responses of clutter, both of which mighthave filtered through as targets during the detection phase. Frequency agility, Doppler, andpolarimetric processing have been used in the discrimination phase. Target recognitionpertains to determination whether the detected target belongs to a class of objects thatthe radar operator is looking for. For example, ballistic missile reentry vehicles and decoysform a class of potential radar objects of interest for detection, while discriminationbetween the two objects in the presence of clutter belongs to the second phase, and theclassification of the missile using features such as nose-tip scattering and material coatingsbelongs to the recognition phase. In this section, essential polarization behavior ofdifferent radar objects such as targets, land and sea clutter are described.13.4.1 TargetsHigh-frequency backscattering from complex targets is essentially a local phenomenon,such as specular scattering or edge diffraction. In fact, mathematical analyses and measurementsshow that the backscattered wave can be essentially regarded as the coherentsum of contributions arising from a finite number of discrete scattering centers [44]. Thewave scattered by each such point source depends on the shape and constitutive materialproperties of a small surface around the scattering center. For example, scattering centerscan be associated with specular reflection off a large convex surface and diffraction arisingfrom discontinuities (e.g., base edge of a cone) and changes in curvature of the body. Theposition of the backscattering centers is sensitive to target aspect, especially in the case ofspecular reflection. Relative motion of the scattering centers also affects the phase of theconstituent waves in the coherent summation.