Multiscale Image Analysis with Stochastic Texture Differences - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2014

Multiscale Image Analysis with Stochastic Texture Differences

Abstract

We introduce a feature descriptor based on stochastic differences between random paths in spatially extended data, like an image. This descriptor behaves like a "texture gradient" and it is sensitive to both spatial and data scales. Applications to inferring characteristic scales in remotely sensed data is presented, together with multi-scale image analysis, and color edge-detection. The method is generic and can be applied to other types of vector or structured data.
presentation_stochadiff.pdf (3.99 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-01120159 , version 1 (23-03-2015)

Identifiers

  • HAL Id : hal-01120159 , version 1

Cite

Nicolas Brodu, Hussein Yahia. Multiscale Image Analysis with Stochastic Texture Differences. Recent Advances in Electronics and Computer Engineering, Feb 2015, Roorkee, India. ⟨hal-01120159⟩
83 View
23 Download

Share

Gmail Facebook X LinkedIn More