Abstract

This paper presents a system for optical character recognition (OCR) that is both powerful and portable, able to analyse images and graphics taken with a camera and incorporated into text texts. Text section extraction and skew correction are the first steps in the procedure. The defined regions are next subjected to binarization, which separates them into lines and characters in preparation for additional examination. After that, these characters are sent to a recognition module. Extensive testing using a dataset of 100 cell phone-taken business card photographs produced an astounding maximum accuracy of 92.74%. An analysis conducted in comparison with the open-source Tesseract OCR engine, which is commonly used on desktop computers, demonstrates the excellent accuracy of the suggested solution. Most notably, the technique uses less memory and shows off its processing power, making it especially appropriate for mobile devices. This invention is very different from traditional OCR research, which is mostly focused on document images scanned by cumbersome desktop scanners with flatbeds that are connected to computers, making them impractical for mobile applications.