Riemannian Clustering of PolSAR Data using the Polar Decomposition - Archive ouverte HAL
Pré-Publication, Document De Travail (Working Paper) Année : 2022

Riemannian Clustering of PolSAR Data using the Polar Decomposition

Résumé

In this manuscript we propose an algorithm for unsupervised classification of PolSAR data, on the manifold of Hermitian positive definite matrices obtained from the polar decomposition of the scattering matrix. The method uses a geodetic metric for evaluating similarity of Hermitian matrices and performs unsupervised classification for both coherent and incoherent targets. Monostatic, full-polarimetric, real and simulated datasets are used for testing the proposed method. With Gaussian clutter, the technique is able to retrieve classification maps similar to those obtained using the standard Wishart algorithm. A refinement of classification results is shown for a simulated dataset with 4 classess. While the Wishart classifier attains an average class accuracy of almost 97%, the proposed method reaches almost 99%. For real PolSAR data, the final classification better preserves the texture information of the original image. As a result, an improved separation is shown between nearby areas of lower intensity, as for example vegetation fields.
Fichier principal
Vignette du fichier
Polar_Riemann_v2_4.pdf (21.8 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03839678 , version 1 (04-11-2022)
hal-03839678 , version 2 (11-04-2023)
hal-03839678 , version 3 (10-11-2023)

Identifiants

  • HAL Id : hal-03839678 , version 1

Citer

Madalina Ciuca, Gabriel Vasile, Marco Congedo, Michel Gay. Riemannian Clustering of PolSAR Data using the Polar Decomposition. 2022. ⟨hal-03839678v1⟩
432 Consultations
68 Téléchargements

Partager

More