سال انتشار: ۱۳۹۰

محل انتشار: پنجمین کنفرانس بین المللی پیشرفتهای علوم و تکنولوژی

تعداد صفحات: ۶

نویسنده(ها):

Mozhgan Fardinpour – Department of Computer, Behbahan Branch, Islamic Azad University, Behbahan, Iran
Navid Samimi Behbahan – Department of Computer, Behbahan Branch, Islamic Azad University, Behbahan, Iran
Zahra Eram – Department of Mathematics, Behbahan Branch, Islamic Azad University, Behbahan, Iran

چکیده:

Making a useful model to recognize handwritten text always has been important for researchers. In the facing research, Farsi handwritten digits which are available in the Hoda dataset, which is a proof for researchers, have been investigated. This method consists of thinning of digits, selecting a number of fixed pixels with equal interval (converting digit to points) on each sample and at last overlapping of all samples which are in one class. The purpose of this research is creating a reference pattern for each digit. In order to recognize the related class of the new sample only we need to put the sample on al ten references to specify the new sample. The results signify a high performance of the method