Open AccessThe need to translate paper documents into electronic form demanded a development of methods and algorithms for automatic processing systems and web publishing unformatted graphic documents of on-line libraries. Translation of scanned images into modern formats of electronic documents using OCR programmes faces serious difficulties. These difficulties are connected with the standardization set of fonts and design of printed documents. There is also a need to maintain the original form of electronic format of such documents. The article discusses the possibility for building an extensible adaptive dictionary of graphic objects, which constitute unformatted graphics documents. Dictionary automatically adjusted as graphics processing and accumulation of statistical information for each new document. This adaptive extensible dictionary of graphic letters, fonts, and other objects of automated particular document processing is called "auto-adaptive font", and a set of its application methods is named "auto-adaptive font technology."
Based on the theory of estimation algorithms, a mathematical model is designed. It allows us to represent all objects of unformatted graphic document in a unified manner to build a feature vector for each object, and evaluate a similarity of these objects in the selected metric. The algorithm of the adaptive models of graphic images is developed and a criterion for combining similar properties in one element to build an auto-adaptive font is offered thus allowing us to build a software core of hardware-software complex for processing the unformatted graphic documents. A standard block diagram of hardware-software complex is developed to process the unformatted graphic documents. The article presents a description of all the blocks of this complex, including document processing station and its interaction with the web server of publishing electronic documents.