Detail of publication

Abstract

Thinking of the motivation for using smart artificial systems to help or even substitute people when dealing with a specific task, one of the first aspects that comes on my mind is the speed computers are capable of working at. These systems are often solemnly called ”artificial intelligence”, which is wrong in my point of view, as they cannot perform the complex cognitive behavior as humans can do. However, when a specific (not complex) behavior is required to deal with a certain task, machines can be effectively used by taking advantage of their working pace and moreover of the fact they never get bored or tired. One of such applications is image processing in general. No living human is capable of browsing and processing thousands of images per second, but a well-designed artificial system possibly could be. This work is focused on reading text in images, especially in scanned documents, and motivated by a call for digitization of huge amount of archival scans and searching in them. The desired functionality of the developed system is what the commonly known shortcut CTRL+F does, however, applied on a scanned document instead of a digital text.