We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

eess.IV

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Electrical Engineering and Systems Science > Image and Video Processing

Title: A Dual-fusion Semantic Segmentation Framework With GAN For SAR Images

Abstract: Deep learning based semantic segmentation is one of the popular methods in remote sensing image segmentation. In this paper, a network based on the widely used encoderdecoder architecture is proposed to accomplish the synthetic aperture radar (SAR) images segmentation. With the better representation capability of optical images, we propose to enrich SAR images with generated optical images via the generative adversative network (GAN) trained by numerous SAR and optical images. These optical images can be used as expansions of original SAR images, thus ensuring robust result of segmentation. Then the optical images generated by the GAN are stitched together with the corresponding real images. An attention module following the stitched data is used to strengthen the representation of the objects. Experiments indicate that our method is efficient compared to other commonly used methods
Comments: 4 pages,4 figures, 2022 IEEE International Geoscience and Remote Sensing Symposium
Subjects: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
Report number: 1052
Cite as: arXiv:2206.01096 [eess.IV]
  (or arXiv:2206.01096v1 [eess.IV] for this version)

Submission history

From: Donghui Lee [view email]
[v1] Thu, 2 Jun 2022 15:22:29 GMT (299kb)

Link back to: arXiv, form interface, contact.