Study of Azerbaijani Hand-Printed Characters Recognition System by New Feature Class and SVM Method`

Abstract

Although there are widely spread Latin scripts in Azerbaijani alphabet, intending special symbols and morphological content of the language requires individual approach for character recognition. In this paper, "soft" (close to human mind, constructed on base of characteristics used in alphabet learning) features and SVM for recognition of Azerbaijani hand-printed characters are used. For character classification, bootstrap resampling procedure of support vector machines is used. Results are compared with results of other feature classes and methods.

Parallel Solution of Features Subset Selection Process for Hand-Printed Character Recognition

“…For this study, we analyzed the data collected from forms filled by different people with different writing skills with Azerbaijani hand-printed characters, digits and special symbols. Classification of the training database was realized by the SVM method, which gives more accurate results for the investigated problem (Ismayilov, 2018).…”