Artificial Neural Network for Identification of Large Letters of the Alphabet
P. Berk, P. Vindiš, M. Lakota, Matjaž Ocepek, Jurij Rakun · DAAAM International Vienna, Vienna 2011 eBooks · 2011
One of the classical artificial neural network (ANN) applications is a system for character recognition.This system is the basis for different types of applications in various institutions, many applications are used in our everyday lives.Therefore, in our application we decided for the synthesis of ANN through the use of backpropagation method with adaptive learning rate and boot function.Character recognition system we developed using Matlab 2008b software tool.Each letter of the alphabet were presented with an array of dimensions [20x20] in the system and created a neural network algorithm is a matrix with a binary input code size [20x20].The result of processing ANN is a useful way of logical value 1, for successfully recognized symbol (letter X).