Real-Time Page Extraction for Document Digitization
Dmytro Darahan, Artur Kovernikov, Oleksii Liashenko, Andriy Lysenko · 2022
The paper analyzes existing methods of processing photo and video images to solve the problem of extracting pages from a video stream. The concept of an algorithm for video segmentation for use in real-time optical character recognition (OCR) tasks is proposed. Possibilities for implementation of the approach, potential shortcomings, and challenges, as well as ways to solve them, are described. The algorithm can become the basis for other algorithms for digitization and analysis of printed and handwritten texts, which can later be used for future research in the field. The novelty of the approach is the use of video semantics algorithms to solve the topical question of text recognition.