Polarimetric radar image classi cation using directional diffusion and descriptive statistics

Romulus Terebes 1 Raul Malutan 1 Monica Borda 1 Christian Germain 2 Lionel Bombrun 2 Ioana Ilea 2, 1
1 Technical University of Cluj-Napoca
2 IMS - Laboratoire de l'intégration, du matériau au système
Abstract : This paper proposes a novel denoising method for polarimetric synthetic aperture radar (PolSAR) image preprocessing tasks with integration and applications for texture-like land images classification tasks. The method is developed using the partial differential equations framework and employs a multi-polarimetric tensor to capture the geometry of fully polarimetric PolSAR data. The filtering intensity is modulated by the multiplicative gradient norm computed on the total scattered power. The method has good texture preservation properties and it is integrated on a PolSAR image classification chain providing good recognition accuracy rates. Visual results on real PolSAR data of maritime pine forests stands are also provided for showing its effectiveness.
Keywords : Radar Polarimetry Partial differential equations Gray level co-occurrence matrix Classification
Document type :
Journal articles
Complete list of metadatas
https://hal.archives-ouvertes.fr/hal-01930194
Contributor : Lionel Bombrun <>
Submitted on : Wednesday, November 21, 2018 - 4:33:37 PM
Last modification on : Monday, November 26, 2018 - 1:30:03 PM

Identifiers

  • HAL Id : hal-01930194, version 1

Collections

CNRS | IMS-BORDEAUX-FUSION | IMS-BORDEAUX

Citation

Romulus Terebes, Raul Malutan, Monica Borda, Christian Germain, Lionel Bombrun, et al.. Polarimetric radar image classi cation using directional diffusion and descriptive statistics. Rev. Roum. Sci. Techn. – Électrotechn. et Énerg, 2018, 63 (1), pp.83-88. ⟨hal-01930194⟩

Export

BibTeX TEI DC DCterms EndNote

Share

Metrics

Record views

32

Contact

CCSD