English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Speckle fluctuations seriously limit the interpretability of syntheticaperture radar (SAR) images. Speckle reduction has thus been the subject ofnumerous works spanning at least four decades. Techniques based on deep neuralnetworks have recently achieved a new level of performance in terms of SARimage restoration quality. Beyond the design of suitable network architecturesor the selection of adequate loss functions, the construction of training setsis of uttermost importance. So far, most approaches have considered asupervised training strategy: the networks are trained to produce outputs asclose as possible to speckle-free reference images. Speckle-free images aregenerally not available, which requires resorting to natural or optical imagesor the selection of stable areas in long time series to circumvent the lack ofground truth. Self-supervision, on the other hand, avoids the use ofspeckle-free images. We introduce a self-supervised strategy based on theseparation of the real and imaginary parts of single-look complex SAR images,called MERLIN (coMplex sElf-supeRvised despeckLINg), and show that it offers astraightforward way to train all kinds of deep despeckling networks. Networkstrained with MERLIN take into account the spatial correlations due to the SARtransfer function specific to a given sensor and imaging mode. By requiringonly a single image, and possibly exploiting large archives, MERLIN opens thedoor to hassle-free as well as large-scale training of despeckling networks.The code of the trained models is made freely available athttps://gitlab.telecom-paris.fr/RING/MERLIN.

As If by Magic: Self-Supervised Training of Deep Despeckling Networks With MERLIN