An Implementation Framework for Fast Image Processing

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

Abstract

Efficient processing of image data is a key aspect of achieving real-time performance for image and video applications. Here, a biologically inspired novel framework which uses a spiral indexing scheme is used to facilitate fast image processing. In particular we demonstrate the effectiveness of our approach on low-level image operations (convolution) and feature extraction (edge detection). Unlike conventional image addressing schemes where the pixels are indexed using two-dimensional Cartesian coordinates, a spiral addressing scheme enables the pixels to be stored in memory adjacent to their immediate neighbours and indexed as a one-dimensional vector. This permits both efficient traversal of the image structure and efficient application of image processing operators. Performance is evaluated by the application of Laplacian edge detection. The results demonstrate the efficiency of the proposed approach compared with a typical two-dimensional implementation
LanguageEnglish
Title of host publicationUnknown Host Publication
Pages1-4
Number of pages4
Publication statusAccepted/In press - 24 Apr 2017
Event3rd International Conference on Robotics and Vision (ICRV 2017) - Wuhan, China
Duration: 24 Apr 2017 → …
http://www.icrv.org/

Conference

Conference3rd International Conference on Robotics and Vision (ICRV 2017)
Period24/04/17 → …
Internet address

Fingerprint

Image processing
Edge detection
Pixels
Convolution
Feature extraction
Data storage equipment
Processing

Keywords

  • feature extraction
  • image frameworks
  • sqiral image

Cite this

@inproceedings{b74ad8dbf9ed45acb058b478c44410f1,
title = "An Implementation Framework for Fast Image Processing",
abstract = "Efficient processing of image data is a key aspect of achieving real-time performance for image and video applications. Here, a biologically inspired novel framework which uses a spiral indexing scheme is used to facilitate fast image processing. In particular we demonstrate the effectiveness of our approach on low-level image operations (convolution) and feature extraction (edge detection). Unlike conventional image addressing schemes where the pixels are indexed using two-dimensional Cartesian coordinates, a spiral addressing scheme enables the pixels to be stored in memory adjacent to their immediate neighbours and indexed as a one-dimensional vector. This permits both efficient traversal of the image structure and efficient application of image processing operators. Performance is evaluated by the application of Laplacian edge detection. The results demonstrate the efficiency of the proposed approach compared with a typical two-dimensional implementation",
keywords = "feature extraction, image frameworks, sqiral image",
author = "John Fegan and Sonya Coleman and Dermot Kerr and Scotney Bryan",
year = "2017",
month = "4",
day = "24",
language = "English",
pages = "1--4",
booktitle = "Unknown Host Publication",

}

Fegan, J, Coleman, S, Kerr, D & Bryan, S 2017, An Implementation Framework for Fast Image Processing. in Unknown Host Publication. pp. 1-4, 3rd International Conference on Robotics and Vision (ICRV 2017), 24/04/17.

An Implementation Framework for Fast Image Processing. / Fegan, John; Coleman, Sonya; Kerr, Dermot; Bryan, Scotney.

Unknown Host Publication. 2017. p. 1-4.

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

TY - GEN

T1 - An Implementation Framework for Fast Image Processing

AU - Fegan, John

AU - Coleman, Sonya

AU - Kerr, Dermot

AU - Bryan, Scotney

PY - 2017/4/24

Y1 - 2017/4/24

N2 - Efficient processing of image data is a key aspect of achieving real-time performance for image and video applications. Here, a biologically inspired novel framework which uses a spiral indexing scheme is used to facilitate fast image processing. In particular we demonstrate the effectiveness of our approach on low-level image operations (convolution) and feature extraction (edge detection). Unlike conventional image addressing schemes where the pixels are indexed using two-dimensional Cartesian coordinates, a spiral addressing scheme enables the pixels to be stored in memory adjacent to their immediate neighbours and indexed as a one-dimensional vector. This permits both efficient traversal of the image structure and efficient application of image processing operators. Performance is evaluated by the application of Laplacian edge detection. The results demonstrate the efficiency of the proposed approach compared with a typical two-dimensional implementation

AB - Efficient processing of image data is a key aspect of achieving real-time performance for image and video applications. Here, a biologically inspired novel framework which uses a spiral indexing scheme is used to facilitate fast image processing. In particular we demonstrate the effectiveness of our approach on low-level image operations (convolution) and feature extraction (edge detection). Unlike conventional image addressing schemes where the pixels are indexed using two-dimensional Cartesian coordinates, a spiral addressing scheme enables the pixels to be stored in memory adjacent to their immediate neighbours and indexed as a one-dimensional vector. This permits both efficient traversal of the image structure and efficient application of image processing operators. Performance is evaluated by the application of Laplacian edge detection. The results demonstrate the efficiency of the proposed approach compared with a typical two-dimensional implementation

KW - feature extraction

KW - image frameworks

KW - sqiral image

M3 - Conference contribution

SP - 1

EP - 4

BT - Unknown Host Publication

ER -