Gabor feature space diffusion via the minimal weighted area method

Chen Sagiv, Nir A. Sochen, Yehoshua Y. Zeevi

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

5 Scopus citations

Abstract

Gabor feature space is elaborated for representation, processing and segmentation of textured images. As a first step of preprocessing of images represented in this space, we introduce an algorithm for Gabor feature space denoising. It is a geometric-based algorithm that applies diffusion-like equation derived from a minimal weighted area functional, introduced previously and applied in the context of stereo reconstruction models [6,12]. In a previous publication we have already demonstrated how to generalize the intensity-based geodesic active contours model to the Gabor spatial-feature space. This space is represented, via the Beltrami framework, as a 2D Riemannian manifold embedded in a 6D space. In this study we apply the minimal weighted area method to smooth the Gabor space features prior to the application of the geodesic active contour mechanism. We show that this ”Weighted Beltrami” approach preserves edges better than the original Beltrami diffusion. Experimental results of this feature space denoisingpro cess and of the geodesic active contour mechanism applied to the denoised feature space are presented.

Original languageEnglish
Title of host publicationEnergy Minimization Methods in Computer Vision and Pattern Recognition - 3rd International Workshop, EMMCVPR 2001, Proceedings
EditorsAnil K. Jain, Mario Figueiredo, Josiane Zerubia
PublisherSpringer Verlag
Pages621-635
Number of pages15
ISBN (Print)3540425233, 9783540425236
DOIs
StatePublished - 2001
Event3rd International Workshop on Energy Minimization Methods in Computer Vision and Pattern Recognition, EMMCVPR 2001 - Sophia Antipolis, France
Duration: 3 Sep 20015 Sep 2001

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2134
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference3rd International Workshop on Energy Minimization Methods in Computer Vision and Pattern Recognition, EMMCVPR 2001
Country/TerritoryFrance
CitySophia Antipolis
Period3/09/015/09/01

Keywords

  • Anisotropic diffusion
  • Beltrami framework
  • Gabor analysis
  • Geodesic active contours
  • Geometric-based algorithms
  • Image manifolds
  • Minimal weighted area method

Fingerprint

Dive into the research topics of 'Gabor feature space diffusion via the minimal weighted area method'. Together they form a unique fingerprint.

Cite this