Iterative Self-Dual Reconstruction on Radar Image Recovery
Imaging systems as ultrasound, sonar, laser and synthetic aperture radar (SAR) are subjected to speckle noise during image acquisition. Before analyzing these images, it is often necessary to remove the speckle noise using filters. We combine properties of two mathematical morphology filters with speckle statistics to propose a signal-dependent noise filter to multiplicative noise. We describe a multiscale scheme that preserves sharp edges while it smooths homogeneous areas, by combining local statistics with two mathematical morphology filters: the alternating sequential and the self-dual reconstruction algorithms. The experimental results show that the proposed approach is less sensitive to varying window sizes when applied to simulated and real SAR images in comparison with standard filters.
- Research Organization:
- Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United States)
- Sponsoring Organization:
- Computational Research Division
- DOE Contract Number:
- DE-AC02-05CH11231
- OSTI ID:
- 986492
- Report Number(s):
- LBNL-3846E; TRN: US201017%%362
- Resource Relation:
- Conference: IEEE 2009 Workshop on Applications of Computer Vision (WACV)
- Country of Publication:
- United States
- Language:
- English
Similar Records
Iterative image-domain decomposition for dual-energy CT
Relationships between autofocus methods for SAR and self-survey techniques for SONAR. [Synthetic Aperture Radar (SAR)]