Automatic Thresholding of Gray-level Using Multi-stage Approach

  • Authors:
  • Sue Wu;Adnan Amin

  • Affiliations:
  • -;-

  • Venue:
  • ICDAR '03 Proceedings of the Seventh International Conference on Document Analysis and Recognition - Volume 1
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

A multi-stage approach is presented for thresholdingdocument images, along with its application. Theproposed method is based on two stages. Globalthresholding is used in first stage to give a preliminaryresult. A second stage then refines the threshold valuebased on local spatial characteristics of the regionsformed in the first stage. It automatically customizes thethresholding of regions that have specific and consistentcharacteristics but are different to other regions in theimage. This technique works well for both simple images,in which the background and foreground are distinct andseparable, and complex images containing multipleregions with different shading/textures. A typicalapplication is postal envelope analysis. The results ofevaluation show significant improvement compared toseveral other global and local thresholding techniques.