Combined detector of locally-oriented structures and corners in images based on a scale-space tensor representation of local neighborhoods of pixels

  • Authors:

  • Bogusław Cyganek

  • Affiliations:

  • University of Mining and Metallurgy, Department of Electronics, Kraków, Poland

  • Venue:
  • ICCS'03 Proceedings of the 2003 international conference on Computational science: PartII
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

Detection of low-level image features such as edges or corners has been an essential task of image processing for many years. Similarly, detectors of such image features constitute basic building blocks of almost every image processing system. However, today's growing amount of vision applications requires at least twofold research directions: search for detectors that work better than the other, at least for a chosen group of images of interest, and - at the other hand - search for new image features, such as textons or oriented structures of local neighborhoods of pixels. In this paper we present a new approach to the old problem of corner detection, as well as detection of areas in images that can be characterized by the same angular orientation. Both detecting techniques are based on a scale-space tensor representation of local structures, and present computationally attractive image feature detectors.