Related Experiment Video
Updated: Aug 3, 2026
08:25
Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
Published on: May 7, 2019
Natural language morphology integration in off-line Arabic optical text recognition
Slim Kanoun1, Adel M Alimi, Yves Lecourtier
1REsearch Group on Intelligent Machines (REGIM), National School of Engineers, University of Sfax, 3038 Sfax, Tunisia. slim.kanoun@yahoo.fr
Summary
This study introduces the affixal approach for Arabic optical character recognition (OCR). This method uses linguistic morphology to improve word and text recognition accuracy.
Area of Science:
- Natural Language Processing
- Computer Vision
- Computational Linguistics
Background:
- Existing Arabic OCR methods often use post-recognition validation or lexicon-directed recognition with statistical models like Hidden Markov Models (HMM) or N-grams.
- These methods have limitations in fully leveraging Arabic's rich morphology.
Purpose of the Study:
- To introduce a novel linguistic-based approach, the affixal approach, for Arabic word and text image recognition.
- To enhance the accuracy and efficiency of Arabic OCR by integrating morphological analysis directly into the recognition process.
Main Methods:
- The proposed affixal approach utilizes linguistic concepts of Arabic vocabulary, specifically focusing on root-based derivation and morphological characterization.
- It categorizes word hypotheses into derived and non-derived forms and analyzes their morphology.
Main Results:
- The affixal approach simplifies the recognition process by using linguistic knowledge.
- It effectively prepares text hypotheses for subsequent analyses, such as syntactic analysis and sentence filtering.
Conclusions:
- The affixal approach offers a more linguistically informed method for Arabic OCR compared to traditional techniques.
- This morphological analysis improves the robustness and utility of OCR outputs for further linguistic processing.
