Open Access Open Access  Restricted Access Subscription Access

An Effective Binarization Method for Readability Improvement of Stain-Affected (degraded) Palm Leaf and other Types of Manuscripts


Affiliations
1 Department of Computer Science, University of Mumbai, Vidyanagari, Santacruz (E), Mumbai 400 098, India
 

A stain is any colour change perceived as lying within the manuscript fibres. Eliminating stains is considered as more significant, since it leads to improved manuscript image quality and hence improved readability. The communication describes an improved binarization method for removing stains from severely degraded manuscript images where the text is unclear. The method is tested on manuscript images on different media having different scripts: palm leaf (Grantha), rock (Brahmi), and paper (Modi, Newari, Persian and Roman), and Document Image Binarization Contest datasets. In this work, we obtained 66.27%, 92.15%, 97.90%, 56.23%, 78.62% and 98.91% readability for rock, palm leaf and paper manuscripts respectively.

Keywords

Binarization, Degradation, Manuscripts, Readability Improvement, Stains.
User
Notifications
Font Size

Abstract Views: 373

PDF Views: 121




  • An Effective Binarization Method for Readability Improvement of Stain-Affected (degraded) Palm Leaf and other Types of Manuscripts

Abstract Views: 373  |  PDF Views: 121

Authors

Lalit Prakash Saxena
Department of Computer Science, University of Mumbai, Vidyanagari, Santacruz (E), Mumbai 400 098, India

Abstract


A stain is any colour change perceived as lying within the manuscript fibres. Eliminating stains is considered as more significant, since it leads to improved manuscript image quality and hence improved readability. The communication describes an improved binarization method for removing stains from severely degraded manuscript images where the text is unclear. The method is tested on manuscript images on different media having different scripts: palm leaf (Grantha), rock (Brahmi), and paper (Modi, Newari, Persian and Roman), and Document Image Binarization Contest datasets. In this work, we obtained 66.27%, 92.15%, 97.90%, 56.23%, 78.62% and 98.91% readability for rock, palm leaf and paper manuscripts respectively.

Keywords


Binarization, Degradation, Manuscripts, Readability Improvement, Stains.



DOI: https://doi.org/10.18520/cs%2Fv107%2Fi3%2F489-496