To perform text editing in formats that do not provide such a function, pre-processing of images is necessary. Most often, text recognition is performed with the creation of a document structure through an OCR system. When the text is automatically recognized by the system, a document is created into which design styles, fonts, document structure, tables, and illustrations are embedded. For simple documents, articles that do not contain complex layout, illustrations, graphs and drawings, this process produces a very decent result. The document is saved in MS Word format, with which the editor or proofreader (native speaker) subsequently works.
The situation is much more complicated with documents whose structure does not allow it to be automatically recreated in an editable format. An example is a technical instruction with a large number of illustrations with a background and callouts, explanations placed within a graphic element and embedded in its context. For editing purposes, these documents are suitable, but they are categorically not suitable for printing and distribution. Such documents have to be retyped and edited by the editor in the text.
An example of an instruction containing text labels, which, after recognition, becomes practically unusable for the editor. |
