A statistical approach of multiple resolution levels for Canny edge detection

Zuraini Othman, Azizi Abdullah, Anton Satria Prabuwono

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

5 Citations (Scopus)

Abstract

Vision processing needs effective feature detectors to estimate the structure and properties of objects in an image. The best known is Canny edge detection that combine a Gaussian low pass filter for noise reduction and non-maximal suppression and hysteresis threshold for edge localization. A possible problem of this approach is that the threshold values. Applying a single fixed threshold to gradient maxima is not an optimal choice. Thus, Canny uses two thresholds values namely Tlow and Thigh to reduce the number of false positive of pixels that represent significant contours in the image. However, by introducing two fixed threshold values are also not an optimal choice due to high variations in images. In this paper we introduce a method that computes the threshold values from the foreground and background image pixels. According to this method, an image is divided into several blocks using at multiple resolution levels. After that, a sampling approach is used on global and local regions to get the optimal thresholds by selecting the highest between class variance values. We have performed experiments on 200 images from the Berkeley dataset. The results show that the proposed method outperforms Canny that uses two fixed threshold values.

Original languageEnglish
Title of host publicationInternational Conference on Intelligent Systems Design and Applications, ISDA
Pages837-841
Number of pages5
DOIs
Publication statusPublished - 2012
Event2012 12th International Conference on Intelligent Systems Design and Applications, ISDA 2012 - Kochi
Duration: 27 Nov 201229 Nov 2012

Other

Other2012 12th International Conference on Intelligent Systems Design and Applications, ISDA 2012
CityKochi
Period27/11/1229/11/12

Fingerprint

Edge detection
Pixels
Low pass filters
Noise abatement
Hysteresis
Sampling
Detectors
Processing
Experiments

Keywords

  • edge detection
  • multiple resolution
  • sampling approach
  • threshold value

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Science Applications
  • Signal Processing
  • Control and Systems Engineering

Cite this

Othman, Z., Abdullah, A., & Prabuwono, A. S. (2012). A statistical approach of multiple resolution levels for Canny edge detection. In International Conference on Intelligent Systems Design and Applications, ISDA (pp. 837-841). [6416646] https://doi.org/10.1109/ISDA.2012.6416646

A statistical approach of multiple resolution levels for Canny edge detection. / Othman, Zuraini; Abdullah, Azizi; Prabuwono, Anton Satria.

International Conference on Intelligent Systems Design and Applications, ISDA. 2012. p. 837-841 6416646.

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

Othman, Z, Abdullah, A & Prabuwono, AS 2012, A statistical approach of multiple resolution levels for Canny edge detection. in International Conference on Intelligent Systems Design and Applications, ISDA., 6416646, pp. 837-841, 2012 12th International Conference on Intelligent Systems Design and Applications, ISDA 2012, Kochi, 27/11/12. https://doi.org/10.1109/ISDA.2012.6416646
Othman Z, Abdullah A, Prabuwono AS. A statistical approach of multiple resolution levels for Canny edge detection. In International Conference on Intelligent Systems Design and Applications, ISDA. 2012. p. 837-841. 6416646 https://doi.org/10.1109/ISDA.2012.6416646
Othman, Zuraini ; Abdullah, Azizi ; Prabuwono, Anton Satria. / A statistical approach of multiple resolution levels for Canny edge detection. International Conference on Intelligent Systems Design and Applications, ISDA. 2012. pp. 837-841
@inproceedings{097e9116093246b19be318e0bf72fe5f,
title = "A statistical approach of multiple resolution levels for Canny edge detection",
abstract = "Vision processing needs effective feature detectors to estimate the structure and properties of objects in an image. The best known is Canny edge detection that combine a Gaussian low pass filter for noise reduction and non-maximal suppression and hysteresis threshold for edge localization. A possible problem of this approach is that the threshold values. Applying a single fixed threshold to gradient maxima is not an optimal choice. Thus, Canny uses two thresholds values namely Tlow and Thigh to reduce the number of false positive of pixels that represent significant contours in the image. However, by introducing two fixed threshold values are also not an optimal choice due to high variations in images. In this paper we introduce a method that computes the threshold values from the foreground and background image pixels. According to this method, an image is divided into several blocks using at multiple resolution levels. After that, a sampling approach is used on global and local regions to get the optimal thresholds by selecting the highest between class variance values. We have performed experiments on 200 images from the Berkeley dataset. The results show that the proposed method outperforms Canny that uses two fixed threshold values.",
keywords = "edge detection, multiple resolution, sampling approach, threshold value",
author = "Zuraini Othman and Azizi Abdullah and Prabuwono, {Anton Satria}",
year = "2012",
doi = "10.1109/ISDA.2012.6416646",
language = "English",
isbn = "9781467351188",
pages = "837--841",
booktitle = "International Conference on Intelligent Systems Design and Applications, ISDA",

}

TY - GEN

T1 - A statistical approach of multiple resolution levels for Canny edge detection

AU - Othman, Zuraini

AU - Abdullah, Azizi

AU - Prabuwono, Anton Satria

PY - 2012

Y1 - 2012

N2 - Vision processing needs effective feature detectors to estimate the structure and properties of objects in an image. The best known is Canny edge detection that combine a Gaussian low pass filter for noise reduction and non-maximal suppression and hysteresis threshold for edge localization. A possible problem of this approach is that the threshold values. Applying a single fixed threshold to gradient maxima is not an optimal choice. Thus, Canny uses two thresholds values namely Tlow and Thigh to reduce the number of false positive of pixels that represent significant contours in the image. However, by introducing two fixed threshold values are also not an optimal choice due to high variations in images. In this paper we introduce a method that computes the threshold values from the foreground and background image pixels. According to this method, an image is divided into several blocks using at multiple resolution levels. After that, a sampling approach is used on global and local regions to get the optimal thresholds by selecting the highest between class variance values. We have performed experiments on 200 images from the Berkeley dataset. The results show that the proposed method outperforms Canny that uses two fixed threshold values.

AB - Vision processing needs effective feature detectors to estimate the structure and properties of objects in an image. The best known is Canny edge detection that combine a Gaussian low pass filter for noise reduction and non-maximal suppression and hysteresis threshold for edge localization. A possible problem of this approach is that the threshold values. Applying a single fixed threshold to gradient maxima is not an optimal choice. Thus, Canny uses two thresholds values namely Tlow and Thigh to reduce the number of false positive of pixels that represent significant contours in the image. However, by introducing two fixed threshold values are also not an optimal choice due to high variations in images. In this paper we introduce a method that computes the threshold values from the foreground and background image pixels. According to this method, an image is divided into several blocks using at multiple resolution levels. After that, a sampling approach is used on global and local regions to get the optimal thresholds by selecting the highest between class variance values. We have performed experiments on 200 images from the Berkeley dataset. The results show that the proposed method outperforms Canny that uses two fixed threshold values.

KW - edge detection

KW - multiple resolution

KW - sampling approach

KW - threshold value

UR - http://www.scopus.com/inward/record.url?scp=84874391722&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84874391722&partnerID=8YFLogxK

U2 - 10.1109/ISDA.2012.6416646

DO - 10.1109/ISDA.2012.6416646

M3 - Conference contribution

AN - SCOPUS:84874391722

SN - 9781467351188

SP - 837

EP - 841

BT - International Conference on Intelligent Systems Design and Applications, ISDA

ER -