Segregation of Text from Images and Translation Using Image Processing and OCR Techniques

Abstract

In today’s multilingual global environment, real-time text extraction and translation from images is crucial for enhancing communication and accessibility. This paper presents a mobile application that enables users to capture or upload images containing foreign text and receive real-time translation into their preferred language. The system integrates Optical Character Recognition (OCR) using Tesseract and image preprocessing techniques via OpenCV to detect and recognize text from complex image environments. Subsequently, the recognized text is translated using Google Translate API. The application is particularly beneficial for tourists, individuals encountering unfamiliar scripts, and use-cases involving signboards, banners, or documents in various languages. Comprehensive experiments were conducted to evaluate recognition performance under different conditions such as font style, size, and color. Results indicate an average recognition accuracy of 87.456%, with English achieving the highest success rate. This system proves effective across multiple languages and environments, offering a robust, user-friendly solution for seamless text translation from images. Future enhancements include improved multilingual support and real-time voice output