Abstract

This study is dealt with Turkish handwritten touching-sloping text recognition. The difficulty of handwritten recognition depends on changing of handwritten person by person and touching-sloping written characters. Also, agglutinative word structure of Turkish language increases difficulty of recognition. It was used lowercase handwritten for recognition system. It was used k-NN for character recognition stage. Character segmentation and lexicon were used together for word recognition. It was blocked choosing incorrect letters using lexicon and corrected recognition of incorrect words. In the study, while performance of character recognition was obtained 90.5%, performance of word recognition was obtained 84%. The lower value of performance of word recognition obtained depends on restricted word in lexicon used for the study.