Title :
Assessment and Estimation of the RVoG Model in Polarimetric SAR Interferometry
Author :
Lopez-Martinez, Carlos ; Alonso-Gonzalez, Alberto
Author_Institution :
Dept. of Signal Theor. & Commun., Tech. Univ. of Catalonia (UPC), Barcelona, Spain
Abstract :
This paper investigates the validity of the random-volume-over-ground (RVoG) scattering model assumption for forest scattering on polarimetric interferometric synthetic aperture radar (PolInSAR) data. The model makes some assumptions about the data and the structure of coherency matrices, namely, the equality of the polarimetric covariance matrices and the affine equivalence of the contracted polarimetric interferometric covariance matrix with a Hermitian matrix. The proposed methodology is divided into two main steps. First, invertible affine transforms (ATs) are studied and proposed as a tool to operate with coherence regions. Based on this analysis, the concept of the trace matrix is introduced as its rank depends on the RVoG model assumption validity. Then, with the objective to consider the effects of speckle noise, we consider a maximum-likelihood (ML) framework, on the hypothesis of data distributed according to the complex Gaussian distribution. Hence, we define the ML estimator (MLE) of the PolInSAR coherency matrix according to the RVoG model assumption and the generalized likelihood ratio test of the model. The validity tests and the MLE are analyzed in terms of simulated and real PolInSAR data, considering P-band and L-band data over tropical and boreal forests.
Keywords :
Gaussian distribution; Hermitian matrices; affine transforms; covariance matrices; geophysical techniques; maximum likelihood estimation; radar interferometry; radar polarimetry; synthetic aperture radar; vegetation; Gaussian distribution; Hermitian matrix; L-band data; ML estimator; MLE; P-band data; PolInSAR coherency matrix; PolInSAR data; RVoG model assessment; RVoG model estimation; affine equivalence; boreal forest; coherence regions; coherency matrix structure; forest scattering; generalized likelihood ratio test; invertible affine transforms; maximum-likelihood framework; polarimetric SAR interferometry; polarimetric interferometric covariance matrix; polarimetric interferometric synthetic aperture radar; random-volume-over-ground scattering model; speckle noise effects; trace matrix; tropical forest; Coherence; Estimation; Matrix decomposition; Speckle; Synthetic aperture radar; Transforms; Affine transform (AT); SAR interferometry; SAR polarimetry; forestry; maximum likelihood (ML); polarimetric synthetic aperture radar interferometry (PolInSAR);
Journal_Title :
Geoscience and Remote Sensing, IEEE Transactions on
DOI :
10.1109/TGRS.2013.2269614