Research on Text Recognition of Natural Scenes for Complex Situations

Abstract

As a great invention that affects human development and progress, writing plays an essential role in human life and learning and the inheritance and development of culture. For thousands of years, the inheritance of Chinese culture has been inseparable from the writing records. Papermaking and printing have played a crucial role in the development of the times. Text extraction and recognition technology in natural scenes, that is, text recognition from photos or videos, has become a popular research direction in computer vision in recent years. Scene text recognition is a fundamental task in computer vision. It is often a critical step in many text-related tasks, such as document retrieval, card recognition, license plate recognition, and other tasks. The vital difference between the text recognition process and text detection is that the text in the photo or video needs to be recognized. Detection and recognition are mainly divided into two steps: locating text areas in photos or videos, that is, finding words or text lines of the bounding box; then the positioned text is recognized. Combining these two steps results in End-to-end Recognition of text. The continuous development of scene text detection and recognition algorithm system will lay the foundation for the research of recognition problems such as multilingual recognition of scene text and formula recognition.