A Study into Entropy-Based Thresholding for Image Edge Detection

AuthorSearch for: ; Search for:
TypeArticle
ConferenceProceedings of Vision Interface '95, May 16-19, 1995., Québec, Quebec
Subjectcomputer vision; entropy; thresholding; segmentation
AbstractAutomated thresholding approaches have normally been applied to gray-level intensity images to differentiate between objects and the background in the image. This paper investigates the use of an entropy-based thresholding approach for determining a reasonable threshold value for an intensity gradient in an edge tracking algorithm and a threshold value for the lengths of edges extracted from an image. The histograms of the intensity gradient of an edge and the lengths of edges generally peak very quickly at low values and quickly drop as their values increase. The entropy-based thresholding technique is adequate for determining a reasonable threshold value for these type of histograms, particularly since it computes the point at which the information content of the two sides of the histogram is a maximum. The paper also demonstrates the importance of reapplying the threshold determination algorithm on different parts of the image, since the threshold value is relative to the distribution in a region of interest. The effects of sparse data on the computation of the threshold are investigated and an example is presented demonstrated the strong impact that sparse data can have.
Publication date
LanguageEnglish
AffiliationNRC Institute for Information Technology; National Research Council Canada
Peer reviewedNo
NRC number38362
NPARC number8913768
Export citationExport as RIS
Report a correctionReport a correction
Record identifierbf0bb73a-03b2-40c2-98d3-25bc99adc07a
Record created2009-04-22
Record modified2016-05-09
Bookmark and share
  • Share this page with Facebook (Opens in a new window)
  • Share this page with Twitter (Opens in a new window)
  • Share this page with Google+ (Opens in a new window)
  • Share this page with Delicious (Opens in a new window)