Year: 2024
Type: article
Abstract: Optical Character Recognition (OCR) algorithm is a technology that converts text images from paper documents into a digital format using electronic devices such as scanners and digital cameras. This process transforms the captured text images into editable and searchable versions using text recognition technology. As advancements in deep learning, AI models have increasingly become pivotal in applications requiring operation on mobile devices without network connectivity, including small underwater devices, high-altitude environments, and license plate recognition systems in front-end cameras. Despite the maturity of general OCR models, there is a notable scarcity of OCR algorithms that are compatible with embedded single-chip microcomputers. These models, capable of functioning autonomously at the front-end without network support, are particularly crucial for remote applications. However, virtually no models for single-chip systems currently support the recognition of the Mongolian language. This study focuses on the development of an OCR system designed for single-chip microcomputers operating without network connectivity. The system is engineered to perform character recognition for Mongolian, English, and Chinese scripts, thereby expanding the utility of front-end single-chip devices. Specifically, the research introduces a novel approach to the recognition of modern Mongolian characters, broadening the scope of OCR system in linguistically diverse contexts.
Source: Data Intelligence
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4402324946