Intra-Version Comparative Performance Analysis of Tesseract on Different Hardware Platforms

The growing demand for document digitisation stems from its ability to transform physical records into accessible, searchable digital formats that enhance preservation, storage efficiency, and retrieval capabilities across diverse organisational contexts. Optical Character Recognition (OCR) is central to converting these records into machine-readable text. Among available open-source OCR solutions, Tesseract has emerged as the predominant choice due to its robust community support and widespread application integration, making it particularly valuable for assistive technology implementations. Comparative analysis between Tesseract versions 4 and 5 reveals significant performance variations when deployed across different hardware architectures, specifically Intel i7 processors versus Raspberry Pi platforms. The evaluation methodology incorporates datasets captured through the Divyanayan scanning device, encompassing both single-column and double-column document layouts in English and Hindi languages to assess recognition capabilities across varying textual complexities and script types. Performance metrics, including character error rate, word error rate, processing time, and time-per-character calculations, provide quantitative frameworks for determining optimal OCR deployment strategies in practical reading machine applications, particularly where hardware constraints and multilingual processing requirements intersect with assistive technology needs for visually impaired users.