Robust TomoSAR focusing for forest height retrieval

Hossein Aghababaee, Giampaolo Ferraioli, Vito Pascazio, Gilda Schirinzi

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

Abstract

This paper aims to discuss and present analysis on robust reconstruction of vertical structure of forested area suing multi-baseline synthetic aperture radar (SAR) data. To deal with problem of low signal to noise (SNR) ratio, robust non-local (NL) techniques of covariance matrix estimation are employed and compared with classical multi-looking approaches. The analysis will consider the quality of the vertical structure profile in such a way to cover the reconstruction quality using both single and fully polarimetric MB data sets. To evaluate the three-dimensional imaging of volumetric media in case of fully polarimetric images, the sum of Kronecker product (SKP) of covariance matrix in which polarization is considered as a way to discriminate the vertically aligned scatterers. From the experimental results, the impact of NL neighborhoods in robust estimation of covariance matrix to resolve the interference signals (e.g. layover) is the most relevant conclusion of the paper.

Original languageEnglish
Title of host publication2018 IEEE International Conference on Environmental Engineering, EE 2018 - Proceedings
Place of PublicationMilan
PublisherIEEE
Pages1-5
Number of pages5
ISBN (Electronic)9781538641828
DOIs
Publication statusPublished - 13 Jun 2018
Externally publishedYes
Event2018 IEEE International Conference on Environmental Engineering, EE 2018 - Milan, Italy
Duration: 12 Mar 201814 Mar 2018

Publication series

Name2018 IEEE International Conference on Environmental Engineering, EE 2018 - Proceedings

Conference

Conference2018 IEEE International Conference on Environmental Engineering, EE 2018
CountryItaly
CityMilan
Period12/03/1814/03/18

Keywords

  • Covariance matrix
  • NL-SAR
  • SAR tomography
  • Vertical structure of forest
  • ITC-CV

Fingerprint Dive into the research topics of 'Robust TomoSAR focusing for forest height retrieval'. Together they form a unique fingerprint.

Cite this