BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY: Image denoising: the man machine contest - Jean-Michel Morel (Eco
 le Normale Supérieure Paris-Saclay)
DTSTART:20190207T150000Z
DTEND:20190207T160000Z
UID:TALK117472@talks.cam.ac.uk
CONTACT:Carola-Bibiane Schoenlieb
DESCRIPTION:This talk addresses one  of the fundamental problems of  signa
 l and image processing\, the  separation  of noise and  signal.  It was  a
 lready the  key problem of  Shannon's foundational Mathematical  Theory of
   Communication. This problem has uncountable applications for image forma
 tion\,  image and video post-production\,  and  feature detection. Since t
 he 70s\, several denoising approaches have been identified and can be grou
 ped in a handful of useful 'denoising principles' with notable progress\, 
  from Fourier analysis\, wavelet theory to sparse decompositions and non-l
 ocal methods. In 2016\,  neural denoisers have started outperforming  (sli
 ghtly) human made  denoising  algorithms. Their principles are quite diffe
 rent. Human  algorithms adopt mathematical assumptions about the image str
 ucture to denoise them.  Neural algorithms learn image structure from vast
  collections of images.  \n We  will explain the  classical theories\,  th
 e neural devices and demonstrate what perspectives and cross-fertilization
   this comparison  yields.  \n\nJoint work with Pablo Arias and Gabriele F
 acciolo
LOCATION:MR 14
END:VEVENT
END:VCALENDAR