Color Edge Detection Hardware based on Geometric Algebra.


Mishra, Biswajit and Wilson, Peter (2006) Color Edge Detection Hardware based on Geometric Algebra. At The 2nd Institution of Engineering and Technology Multimedia Conference 2006, London, UK, 29 - 30 Nov 2006.

This is the latest version of this item.

Download

[img] PDF
Download (767Kb)

Description/Abstract

Modern techniques treat color images as separate monochrome images for processing. Partly, because there is no straightforward generalization of linear filters available for color. However the algorithms yield more accurate results when the correlation among color bands are exploited which shows fundamental difference to process the color images. Earlier work[8] reported the transformation of the color images using Quaternion Fourier Transforms and the realization of a holistic filter based on Quaternion convolution. Here, we discuss the rotor based convolution techniques, a generalization of the previous work, within a new mathematical framework in Geometric Algebra. Based thereupon, a novel hardware architecture is proposed. Experiments show the edge detection with this technique belong to a class of linear vector filter and is holistic in nature. It is tailored for image processing applications, providing an acceptable application performance requirements. The usefulness of the introduced approach was demonstrated by analyzing and implementing a computation intensive edge detection algorithm on this hardware.

Item Type: Conference or Workshop Item (Poster)
Additional Information: Event Dates: 29-30 Nov 2006
Keywords: Geometric Algebra, Color Edge detection, Rotor Convolution, Hardware
Divisions: Faculty of Physical Sciences and Engineering > Electronics and Computer Science > EEE
ePrint ID: 263188
Date Deposited: 16 Nov 2006
Last Modified: 27 Mar 2014 20:06
Further Information:Google Scholar
URI: http://eprints.soton.ac.uk/id/eprint/263188

Available Versions of this Item

  • Color Edge Detection Hardware based on Geometric Algebra. (deposited 16 Nov 2006) [Currently Displayed]

Actions (login required)

View Item View Item

Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.

View more statistics