Bowstring-based dual-threshold computation method for adaptive Canny edge detector

Xiangdong Liu, Yin Yu, Bing Liu, Zhen Li

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

10 Citations (Scopus)

Abstract

This paper proposed a novel dual-threshold computation method of Canny edge detector based on gradient magnitude histogram (GMH), targeting with the adaptive acquisition of low-/high-threshold for unimodal hysteresis thresholding. With the introduction of the bowstring concept, which accurately measures the tendency of the GMH on the whole, the dual-threshold computation is implemented by adaptive-searching two tangent points with transitional characteristics. This skillful algorithm of the dual-threshold computation method is further evaluated by using the receiver operating characteristics (ROC) curve evaluation method. The detailed comparison to the Otsu's method is presented and demonstrates the reliability and robust performance of the proposed dual-threshold computation method.

Original languageEnglish
Title of host publicationProceedings of 2013 28th International Conference on Image and Vision Computing New Zealand, IVCNZ 2013
Pages13-18
Number of pages6
DOIs
Publication statusPublished - 2013
Event2013 28th International Conference on Image and Vision Computing New Zealand, IVCNZ 2013 - Wellington, New Zealand
Duration: 27 Nov 201329 Nov 2013

Publication series

NameInternational Conference Image and Vision Computing New Zealand
ISSN (Print)2151-2191
ISSN (Electronic)2151-2205

Conference

Conference2013 28th International Conference on Image and Vision Computing New Zealand, IVCNZ 2013
Country/TerritoryNew Zealand
CityWellington
Period27/11/1329/11/13

Keywords

  • Adaptive dual-threshold computation
  • Canny edge detector
  • Gradient magnitude histogram (GMH)
  • Unsupervised hysteresis thresholding

Fingerprint

Dive into the research topics of 'Bowstring-based dual-threshold computation method for adaptive Canny edge detector'. Together they form a unique fingerprint.

Cite this