BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Modular Optimization Strategies for Compression and Restoration of
  Visual Data - Yehuda Dar (Technion\, Haifa)
DTSTART:20190520T140000Z
DTEND:20190520T150000Z
UID:TALK125197@talks.cam.ac.uk
CONTACT:Carola-Bibiane Schoenlieb
DESCRIPTION:Many important imaging problems have complex structures that c
 hallenge the design of their algorithmic solutions. Contemporary methods f
 or addressing such intricate problems are often formed as conceptually-sim
 ple iterative procedures that rely on well-engineered black-box modules\, 
 usually in the form of existing denoising techniques. \nThis talk overview
 s an alternative optimization framework to address intricate imaging probl
 ems based on prominent compression standards as the fundamental modules it
 eratively applied. The proposed algorithms utilize the alternating directi
 on method of multipliers (ADMM) optimization technique to address seemingl
 y impractical tasks\, involving aspects of acquisition\, compression and r
 endering of images. Specifically\, we demonstrate how the state-of-the-art
  HEVC compression standard can be adjusted to several layouts of acquisiti
 on and rendering systems\, such that the end-to-end reconstruction errors 
 of the systems are optimized with respect to the compression bit-costs. Th
 e experiments established our method as the best approach for optimizing t
 he system performance at high bit-rates from the compression standpoint.\n
 Moreover\, we relate the proposed approach also to image restoration using
  complexity regularization\, where the likelihood of candidate solutions i
 s evaluated based on their compression bit-costs. Our proposed methods res
 tore signals by leveraging state-of-the-art models designed for compressio
 n. The presented experiments show good results for image deblurring and in
 painting using the JPEG2000 and HEVC compression standards.\n\nThis is joi
 nt work with Prof. Alfred Bruckstein and Prof. Michael Elad.
LOCATION:MR 4\, Centre for Mathematical Sciences
END:VEVENT
END:VCALENDAR