دانلود مقاله Providing a mechanism for detecting Persian and Arabic handwriting using networks Kohonen
سال انتشار: ۱۳۹۱
محل انتشار: دومین کنفرانس ملی مهندسی نرم افزار
تعداد صفحات: ۴
Farnaz Safaee Nik – M.A. Computer Architecture Systems Student
In this article we follow goals Considering a new method for displaying online Farsi and Arabic characters through approaching X polynomial peculiarities and registered Y fromonline preprocessing. Designing a classification from Kohnen nerve net that could apply this method for displaying. The preprocessing phaseincludes characters simulating for sensible noises and also repeated sampling for normalization of the actions regarding to the speed of character writer in order the normalization could be done regarding to the place and scale of the character. We display an online character by extending one polynomial on the stream of each peculiarity from the signal of the character. For achieving multipliers of Coefficients of polynomial equation,we resolve the method of uncertain multipliers after that we arrive to an equation that we resolve the method of the Dotil and we can give theCoefficients with this method. We learn Kohonen memory the method of view that weight of the memory wl will be Coefficients of the polynomial equation that we calculated the minimum distance from the block method.