OPEN ACCESS JOURNALS
|JournalsTopicsAuthorsEditor's Choice||About PsyJournals.ruContact Us|
Character Recognition Based on the Markov Chains 226
Yuriev G.A., Programmer, Chair of Applied Computer Informatics, Information Technologies faculty, Moscow State University of Psychology and Education, Moscow, Russia, Moscow, Russia, email@example.com
A fundamentally new algorithm for character recognition is described, which is based on the abilities of the Markov chains [1; 5] – the Markov models with discrete states and discrete time. The applied apparatus is widely used in solving the problems of recognition, but it is noted that traditionally it has been used somewhat differently. The advantage of this method is the high speed of the tuning (teaching), the ability to specify the arbitrary and required reliability of the result and to modify it in the work process of the program system. The algorithm is successfully implemented in the El-Reader [2; 3; 4] – a system of recognition and vocalization of the flat bed texts. The main advantage of it is its resistance to change in the font styles. The prerequisites for the development, theoretical justification and description of the algorithm used in the software implementation are provided in the article. The accurate statistical evaluation of the reliability of recognition for the given parameters is presented. It is emphasized that the algorithm has a number of advantages over traditional approaches, in case of working with the distorted images.
Keywords: Markov chains, recognition, wavelet transformation, image analysis
Column: Interdisciplinary Researches