Optical Character Recognition Accuracy on Degraded Documents

Marc Jordan Ceballos Saladaga; Florence Jean Talirongan

Optical Character Recognition Accuracy on Degraded Documents

Author(s)	Mr. Marc Jordan Ceballos Saladaga, Florence Jean Talirongan
Country	Philippines
Abstract	This paper examines the OCR performance of EasyOCR across four types of physical degradation: crumpled, wet, dirty, and normal documents. Specifically, this work seeks to understand the effect of physical document degradation on OCR performance and to assist in improving digitization workflows for academic and administrative documents. The experimental setup was rigorously developed, starting with preparing a piece of text to be printed, physically degrading it into one of four types, and finally scanning it. The scan is then processed, and OCR is applied. OCR accuracy is calculated at the character and word levels, at a defined rate, and statistics are generated by summarizing the collected data in tables using descriptive statistics. It was found that OCR performance depends largely on the document's physical condition. By comparing other documents relative to the normal one, which achieved an average accuracy of 94.7%, the performance obtained for the crumpled document averaged 86.9%, and for the dirty document, it averaged 80.9%. This clearly shows that wrinkling and smudges distort character shapes and reduce their recognizability. The wet documents had the lowest average accuracy, at 72.7%. This demonstrates that stroke blurring and faded ink are critical factors in OCR failure. It has to be considered that OCR systems can be prone to failure with physically damaged documents, especially when water damage is present, which poses the greatest problem. It is proposed that preliminary preprocessing methods, such as noise reduction, enhancement, and morphological operations, be employed to restore damaged texts and improve OCR accuracy. An adaptive OCR pipeline, its restoration component, and multilingual capacity should be incorporated into the process to increase the efficiency of digitization at the institutional level. By providing deeper insights into OCR systems' susceptibility to document degradation, this work can further strengthen efforts to achieve accurate automated text extraction and support archival purposes, academic information resources, and research on the design of digitization techniques.
Keywords	Optical Character Recognition, EasyOCR, document degradation, accuracy analysis, digitization
Field	Computer > Data / Information
Published In	Volume 17, Issue 2, April-June 2026
Published On	2026-06-02

View / Download PDF File

doi

CrossRef DOI is assigned to each research paper published in our journal.

IJSAT DOI prefix is
10.71097/IJSAT

Downloads

Research Paper Format Copyright Permission Form and Undertaking Form Cover Page Vol 17 Isu 2 Cover Page Vol 17 Isu 1 Cover Page Vol 16 Isu 4

All research papers published on this website are licensed under Creative Commons Attribution-ShareAlike 4.0 International License, and all rights belong to their respective authors/researchers.

CC-BY-SA

About IJSAT Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us	Message on WhatsApp	+91-9687-182-185	editor@ijsat.org

International Journal on Science and Technology

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Bi-monthly Scholarly International Journal

Optical Character Recognition Accuracy on Degraded Documents

Share this