With the help of algorithms based on artificial intelligence, machine learning, and ABBYY’s Adaptive Document Recognition Technology, ABBYY FineReader Engine analyses the document layout and evaluates individual hypothesis.
During the document analysis stage, the document is split into individual pages – the layout of each page is systematically checked to detect placement of text, images, barcodes and table elements. At the same time, the document as an entity is examined and its logical structure is detected. This way, the ‘function’ of text elements is understood – for example, headers and footers will be identified as such and not mistaken for text snippets on individual pages.
Information about text, pictures and formatting elements will be saved - and used later during the final document reconstruction. The result will be an exactly reconstructed document – for example in Word format – including its formatting elements such as tables, fonts and font styles, headers, footers, page numbers, footnotes, table of contents, and others.
AI-based algorithms are as well used to increase the recognition accuracy, especially of the European languages.