A Variational Framework for Active and Adaptative Segmentation of Vector Valued Images

  • Authors:
  • M. Rousson;R. Deriche

  • Affiliations:
  • -;-

  • Venue:
  • MOTION '02 Proceedings of the Workshop on Motion and Video Computing
  • Year:
  • 2002

Quantified Score

Hi-index 0.01

Visualization

Abstract

During the last few years, many efforts have been done inintegrating different informations in a variational frameworkto segment images. Recent works on curve propagationwere able to incorporate stochastic informations [9, 6]and prior knowledge on shapes [3, 7]. The informationinserted in these studies is most of the time extracted offline.Meanwhile, other approaches have proposed to extractregion information during the segmentation processitself [2, 4, 8].Following these new approaches and extending the work in[6] to vector-valued images, we propose in this paper anentirely variational framework to approach the segmentationproblem. Both, the image partition and the statisticalparameters for each region are unkown.After a brief reminder on recent segmenting methods, wewill present a variational formulation obtained from abayesian model. After that, we will show two different differentiationsdriving to the same evolution equations. De-tailedstudies on gray and color images of the 2-phase casewill follow. We will finish on an application to trackingwhich shows benefits of our dynamical framework.