Histogram Peak Ratio-Based Binarization for Historical Document Image

Mahastama, Aditya W. and Krisnawati, Lucia D. (2017) Histogram Peak Ratio-Based Binarization for Historical Document Image. In: Proceedings of 2017 International Conference on Smart Cities, Automation & Intelligent Computing Systems (ICON-SONICS 2017), 08 November 207, Yogyakarta.

Full text not available from this repository.

Abstract

The emergence of large scale digitization projects transforming printed heritage into digitally available resources in Europe and the United States has led to the Digital Renaissance era. The aim of these projects is to preserve the printed cultural heritage and to integrate their intellectual content into the modern information. To achieve this goal, the digitizing process, i.e. transforming a scanned book into an electronic text, becomes necessary. The first step of digitizing process is the preprocessing which involves the segmentation of the foreground, i.e. the text, from the rest of the document. With the goal of digitizing the manuscripts written in Javanese characters, this study proposes a novel approach of foreground segmentation which is intended to serve dual functions, namely to acquire the text characters and also to improve the quality of the document images from their degradation caused by nature or the age. Our method is based on the computation of histogram peak ratio to determine the threshold value of segmentation. Being experimented on Javanese manuscripts in good and degraded conditions, the performance of our method proves to be excellent as its segmentation success rate achieves 100% for manuscripts in good condition. Its performance in segmenting degraded manuscripts caused by holes, sellotape, and bleed-trough effect could be claimed more than satisfying as its success rate achieves 80%.

Item Type: Conference or Workshop Item (Paper)
Subjects: 000 Computer Science, Information and General Works > 000 Computer Science, Knowledge and Systems > 006 Special Computer Methods
900 History and Geography > 900 History > 902 Miscellany of History
Divisions: Universitas Multimedia Nusantara
Depositing User: mr admin umn
Date Deposited: 01 Mar 2018 04:31
Last Modified: 05 Aug 2022 02:46
URI: https://kc.umn.ac.id/id/eprint/2788

Actions (login required)

View Item View Item