The Z/spl pi/M algorithm: a method for interferometric image reconstruction in SAR/SAS

摘要: This paper presents an effective algorithm for absolute phase (not simply modulo-2-/spl pi/) estimation from incomplete, noisy and modulo-2/spl pi/ observations in interferometric aperture radar sonar (InSAR/InSAS). The adopted framework is also representative of other applications such as optical interferometry, magnetic resonance imaging diffraction tomography. Bayesian viewpoint adopted; the observation density 2-/spl pi/-periodic accounts pair decorrelation system noise; a priori probability modeled by compound Gauss-Markov random field (CGMRF) tailored to piecewise smooth images. We propose iterative scheme computation maximum posteriori (MAP) estimate. Each iteration embodies discrete optimization step (Z-step), implemented network programming techniques conditional modes (ICM) (/spl pi/-step). Accordingly, termed Z/spl pi/M, where letter M stands maximization. An important contribution simultaneous implementation unwrapping (inference 2/spl pi/-multiples) smoothing (denoising observations). improves considerably accuracy estimates compared methods which data low-pass filtered prior unwrapping. A set experimental results, comparing proposed with alternative methods, illustrates effectiveness our approach.