Epitomic analysis of appearance and shape

  • Authors:
  • Nebojsa Jojic;Brendan J. Frey;Anitha Kannan

  • Affiliations:
  • -;-;-

  • Venue:
  • ICCV '03 Proceedings of the Ninth IEEE International Conference on Computer Vision - Volume 2
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

We present novel simple appearance and shape models that we callepitomes. The epitome of an image is its miniature, condensedversion containing the essence of the textural and shape propertiesof the image. As opposed to previously used simple image models,such as templates or basis functions, the size of the epitome isconsiderably smaller than the size of the image or object itrepresents, but the epitome still contains most constitute elementsneeded to reconstruct the image (Fig. 1). A collection of imagesoften shares an epitome, e.g., when images are a few consecutiveframes from a video sequence, or when they are photographs ofsimilar objects.A particular image in a collection is defined byits epitome and a smooth mapping from the epitome to the imagepixels. When the epitomic representation is used within ahierarchical generative model, appropriate inference algorithms canbe derived to extract the epitome from a single image or acollection of images and at the same time perform various inferencetasks, such as image segmentation, motion estimation, objectremoval and super-resolution.