Complex Images and Complex Filters: A Unified Model for Encoding and Matching Shape and Colour

  • Authors:
  • Terry Caelli;Andrew McCabe

  • Affiliations:
  • -;-

  • Venue:
  • ICAPR '01 Proceedings of the Second International Conference on Advances in Pattern Recognition
  • Year:
  • 2001

Quantified Score

Hi-index 0.00

Visualization

Abstract

In many practical areas of visual pattern recognition the images are complex-valued. Examples include images generated from 2-dimensional colour, motion, radar and laser sensors. To this date the majority of encoding and matching schemes for such images do not treat the complex nature of the data in a unified way but, rather, separate the associated "channels" and combine them after processing each image attribute. In this paper we describe a technique which utilizes properties of the complex Fourier transform of complex images and develop new types of complex filters for colour and shape specific feature extraction and pattern matching. Results are encouraging particularly under quite noisy conditions.