Iterative PolInSAR Target Decomposition for Scattering Characterization and Building Detection
In densely rotated built-up areas, restricted by interactive and complex scatterings, most polarimetric synthetic aperture radar scattering analyses and unsupervised building detection algorithms have failed, especially under conditions of large radar look angles. To handle this problem, an iterativ...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10938847/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In densely rotated built-up areas, restricted by interactive and complex scatterings, most polarimetric synthetic aperture radar scattering analyses and unsupervised building detection algorithms have failed, especially under conditions of large radar look angles. To handle this problem, an iterative polarimetric interferometric synthetic aperture radar (PolInSAR) target decomposition method for scattering characterization and building detection is proposed in this article, and it consists of three key components. Specifically, by analyzing basic scatterers and electromagnetic wave propagation, the coherent volume scattering is assigned to densely rotated built-up areas. Based on it, a five-component PolInSAR target decomposition method is proposed for unambiguous scattering characterization, where repeat-pass PolInSAR coherence is introduced to aid in unambiguous interpretation by dividing natural areas, nondensely rotated built-up areas, and densely rotated built-up areas. Moreover, to overcome the failure of simple segmentation and deeply explore the scattering differences between densely rotated buildings and forests, an iterative framework integrating self-organizing map (SOM) and PolInSAR target decomposition is finally proposed. SOM uses PolInSAR target decomposition results to refine the segmentation across the three areas, feeding back refined outcomes to the target decomposition module iteratively. This process will ultimately enhance features and improve building detection accuracy. Experiments on three sets of PolInSAR data confirm the validity of the proposed framework, with more reasonable target decomposition results and more accurate building detection results, especially in densely rotated built-up areas. |
|---|---|
| ISSN: | 1939-1404 2151-1535 |