Fast low-level multi-scale feature extraction for hexagonal images

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Inspired by the human vision system and its capability to process in real-time, an efficient framework for low-level feature extraction on hexagonal pixel-based images is presented. This is achieved by utilizing the spiral architecture addressing scheme to simulate eye-tremor along with the convolution of non-overlapping gradient masks. Using sparse spiral convolution and the development of cluster operators, we obtain a set of output image responses “a-trous” that is subsequently collated into a consolidated output response; it is also demonstrated that this framework can be extended to feature extraction at different scales. We show that the proposed framework is considerably faster than using conventional spiral convolution or the use of look-up tables for direct access to hexagonal pixel neighbourhood addresses.
LanguageEnglish
Title of host publicationUnknown Host Publication
Pages342-345
Number of pages4
Publication statusE-pub ahead of print - 20 Jul 2017
Event2017 Fifteenth IAPR International Conference on Machine Vision Applications (MVA) - Japan
Duration: 20 Jul 2017 → …

Conference

Conference2017 Fifteenth IAPR International Conference on Machine Vision Applications (MVA)
Period20/07/17 → …

Fingerprint

Convolution
Feature extraction
Pixels
Mathematical operators
Masks

Keywords

  • Spirals
  • Convolution
  • Machine vision
  • Feature extraction
  • Indexes
  • Computer architecture
  • Organisations

Cite this

@inproceedings{195d38355a08423fbdb5ae4bec3b898b,
title = "Fast low-level multi-scale feature extraction for hexagonal images",
abstract = "Inspired by the human vision system and its capability to process in real-time, an efficient framework for low-level feature extraction on hexagonal pixel-based images is presented. This is achieved by utilizing the spiral architecture addressing scheme to simulate eye-tremor along with the convolution of non-overlapping gradient masks. Using sparse spiral convolution and the development of cluster operators, we obtain a set of output image responses “a-trous” that is subsequently collated into a consolidated output response; it is also demonstrated that this framework can be extended to feature extraction at different scales. We show that the proposed framework is considerably faster than using conventional spiral convolution or the use of look-up tables for direct access to hexagonal pixel neighbourhood addresses.",
keywords = "Spirals, Convolution, Machine vision, Feature extraction, Indexes, Computer architecture, Organisations",
author = "SA Coleman and Scotney Bryan and Bryan Gardiner",
year = "2017",
month = "7",
day = "20",
language = "English",
isbn = "978-4-9011-2216-0",
pages = "342--345",
booktitle = "Unknown Host Publication",

}

Coleman, SA, Bryan, S & Gardiner, B 2017, Fast low-level multi-scale feature extraction for hexagonal images. in Unknown Host Publication. pp. 342-345, 2017 Fifteenth IAPR International Conference on Machine Vision Applications (MVA), 20/07/17.

Fast low-level multi-scale feature extraction for hexagonal images. / Coleman, SA; Bryan, Scotney; Gardiner, Bryan.

Unknown Host Publication. 2017. p. 342-345.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Fast low-level multi-scale feature extraction for hexagonal images

AU - Coleman, SA

AU - Bryan, Scotney

AU - Gardiner, Bryan

PY - 2017/7/20

Y1 - 2017/7/20

N2 - Inspired by the human vision system and its capability to process in real-time, an efficient framework for low-level feature extraction on hexagonal pixel-based images is presented. This is achieved by utilizing the spiral architecture addressing scheme to simulate eye-tremor along with the convolution of non-overlapping gradient masks. Using sparse spiral convolution and the development of cluster operators, we obtain a set of output image responses “a-trous” that is subsequently collated into a consolidated output response; it is also demonstrated that this framework can be extended to feature extraction at different scales. We show that the proposed framework is considerably faster than using conventional spiral convolution or the use of look-up tables for direct access to hexagonal pixel neighbourhood addresses.

AB - Inspired by the human vision system and its capability to process in real-time, an efficient framework for low-level feature extraction on hexagonal pixel-based images is presented. This is achieved by utilizing the spiral architecture addressing scheme to simulate eye-tremor along with the convolution of non-overlapping gradient masks. Using sparse spiral convolution and the development of cluster operators, we obtain a set of output image responses “a-trous” that is subsequently collated into a consolidated output response; it is also demonstrated that this framework can be extended to feature extraction at different scales. We show that the proposed framework is considerably faster than using conventional spiral convolution or the use of look-up tables for direct access to hexagonal pixel neighbourhood addresses.

KW - Spirals

KW - Convolution

KW - Machine vision

KW - Feature extraction

KW - Indexes

KW - Computer architecture

KW - Organisations

M3 - Conference contribution

SN - 978-4-9011-2216-0

SP - 342

EP - 345

BT - Unknown Host Publication

ER -