Work

Enhanced OCR Pipeline for Regional and Historic Records

Year: 2026

Type: article

Abstract: Optical Character Recognition (OCR) plays a pivotal role in digitising historical and regional documents, enabling preservation and accessibility of cultural heritage. This paper presents an Enhanced OCR Pipeline that addresses the challenges of processing degraded, handwritten, and multi-lingual regional records. The proposed system integrates advanced image preprocessing techniques including CLAHE contrast enhancement, bilateral filtering, and Sauvola binarization with intelligent layout analysis, table detection, and multi-format export capabilities. Built on the Tesseract OCR engine, the pipeline supports over 100 languages including RTL scripts and provides per-word confidence scoring with comprehensive quality metrics. Experimental results demonstrate an accuracy of 95%+ on clean modern documents and 85%+ on historical records.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.19346060

Authors: Ameer Akbar, R, Nivin, JK, Sathis Kannan, R, Sudalai Mani, S

Language: English

Topic: Handwritten Text Recognition Techniques

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

OpenAlex ID: W7147436687

FWCI:

Cites:

Cited by:

Related to: