Handwritten Assamese Numeral Recognition

MATLAB Interdisciplinary Research 2022

Project Overview

This project addresses handwritten digit recognition for the Assamese script, exploring how pattern recognition techniques can be adapted for non-Latin writing systems. The work highlights the importance of cultural and linguistic diversity in technology development.

Key Aspects

  • Developed feature extraction methods specific to Assamese numerals
  • Created classification algorithms that recognize distinctive structural elements
  • Built a complete processing pipeline from image acquisition to numeral classification
  • Evaluated performance across different writing styles and image qualities

Cultural Inclusion in Technology

While digit recognition is well-established for Western scripts, many world languages remain underrepresented in technology. This project contributes to more inclusive technology by addressing recognition systems for the Assamese language, spoken by approximately 15 million people in northeastern India.

Methodological Approach

Rather than directly applying techniques optimized for Latin scripts, this project analyzed the unique structural characteristics of Assamese numerals to develop tailored feature extraction methods. This culturally informed approach resulted in higher accuracy and better generalization.

Technologies Used

  • MATLAB for implementation and analysis
  • Custom feature extraction algorithms
  • Statistical pattern recognition methods
  • Performance evaluation frameworks

Phone

Address

Leoben, Austria