Department of Computer Engineering, Omidiyeh Branch, Islamic Azad University, Omidiyeh, Iran
Online published on 7 November, 2013.
In this article, changes in the slope of Persian handwritten digits have been used to describe each digit. The main purpose is to present a method for online rather than offline recognition of handwritten digits. In this method, after several pre-processing steps, the changes of slop of each digit are calculated from the left bottom corner with equal distance, and it is continued along the present curvature. These features are used as the input for HMM algorithm and in order to compare the similarities of each sample with the related references. Finally, the system, designed on HODA database, cited by the researchers, has been applied. The data set includes 1300 samples, that 91. 8% as correct recognition has been obtained.
Persian Handwritten Digits, Hidden Markov Model, Changes in the Slope of Digits, HODA dataset