Initialization of Deep Learning Zero-shot Blind Image Deconvolution

Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21340%2F23%3A00369785" target="_blank" >RIV/68407700:21340/23:00369785 - isvavai.cz</a>
Result on the web
—
DOI - Digital Object Identifier
—

Result language
angličtina
Original language name
Initialization of Deep Learning Zero-shot Blind Image Deconvolution
Original language description
The aim of the blind image deconvolution is to recover a sharp image from a blurred one. Assuming that there is no other data than the one blurred image, the problem is highly ill-posed. Approaches based on bayesian models that attempt to describe an image statistics using priors had played a major role in the zero-shot blind image deblurring until a deep image prior (DIP) and subsequently a DIP framework for the blind image deconvolution were proposed. This method differes from the traditional ones in many aspects, one of them being its initialization. This paper examines influence of the initialization though the modeconnectivity method and discusses a possible role of a power spectra of the sharp image.
Czech name
—
Czech description
—

Publication year
2023
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Article name in the collection
SPMS 2022/23 Stochastic and Physical Monitoring Systems, Proceedings of the international conferences
ISBN
978-80-01-07250-9
ISSN
—
e-ISSN
—
Number of pages
8
Pages from-to
1-8
Publisher name
České vysoké učení technické v Praze
Place of publication
Praha
Event location
Sloup v Čechách
Event date
Jun 26, 2023
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
—

Similar results(10)