Date(s) : 16/05/2013 iCal
14 h 00 min - 15 h 00 min
The Posterior Mean of the Total Variation model\nBy Lionel Moisan\, MAP5.\n\nAbstract:\nThe Total Variation image (or signal) denoising model is a variational approach\nthat can be interpreted\, in a Bayesian framework\, as a search for the maximum\npoint of the posterior density (Maximum A Posteriori estimator). This maximization\naspect is partly responsible for a restoration bias called « staircasing effect »\,\nthat is\, the outbreak of quasi-constant regions separated by sharp edges in the\nintensity map. In this talk\, we discuss a variant of this model that considers\nthe expectation of the posterior distribution instead of its maximum point.\nWe demonstrate in particular\, both theoretically and numerically\, that images\ndenoised with this model do not suffer from the staircasing effect.