Language, script, and font recognition

Abstract

Automatic identification of a language within a text document containing multiple scripts and fonts is a challenging task, as it is not only linked with the shape, size, and style of the characters and symbols used in the formation of the text but also admixed with more crucial factors such as the forms and size of pages, layout of written text, spacing between text lines, design of characters, density of information, directionality of text composition, etc. Therefore, successful management of the various types of information in the act of character, script, and language recognition requires an intelligent system that can elegantly deal with all these factors and issues along with other secondary factors such as language identity, writing system, ethnicity, anthropology, etc. Due to such complexities, identification of script vis-à-vis language has been a real challenge in optical character recognition (OCR) and information retrieval technology. Considering the global upsurge of so-called minor and/or unknown languages, it has become a technological challenge to develop automatic or semiautomatic systems that can identify a language vis-à-vis a script in which a particular piece of text document is composed. Bearing these issues in mind, an attempt is initiated in this chapter to address some of the methods and approaches developed so far for language, script, and font recognition for written text documents. The first section, after presenting a general overview of language, deals with the information about the origin of language, the difficulties faced in language identification, and the existing approaches to language identification. The second section presents an overview of script, differentiates between single- and multiscript documents, describes script identification technologies and the challenges involved therein, focuses on the process of machine-printed script identification, and then addresses the issues involved in handwritten script identification. The third section tries to define font terminologies, addresses the problems involved in font generation, refers to the phenomenon of font variation in a language, and discusses strategies for font and style recognition. Thus, the chapter depicts a panoramic portrait of the three basic components involved in OCR technology: the problems and issues involved, the milestones achieved so far, and the challenges that still lie ahead.