SINOPSIS

Means of off-line recognition of handwritten symbols in a separately-handwritten mathematical text, as well as the structural representation of mathematical formulas and a mathematical text as a whole are proposed. The method of construction of sets of belonging (containers) on the basis of learning samples for classification of handwritten symbols is considered. While recognition of a separately-handwritten mathematical text, we use serial context-sensitive grouping of symbols by specially developed data structures for representation of a text and formulas. These structures assist with the further export of recognized information in electronic documents of different formats. The algorithms of recognition of a separately-handwritten mathematical text and its export into an electronic document of the LaTeX format are developed, the results of a computing experiment are represented as well.