Metode Jaringan Saraf Tiruan Penjejakan Balik Untuk Pengenalan Huruf Cetak Pada Citra Digital

Peneliti : Fauziah, S.Kom, M.MSI, M. Iwan Wahyuddin, ST, MT.

ABSTRACT
Character recognition is a part of computer science to help data processing. One of the
character recognition techniques is the neural network technique. This method is simulating the
principal of human brain consist of neurons to process input using the weight given to produce
output.
The main goal of this experiment is to create software that can recognize printed characters
using neural network methods. The architecture used is the multilayer neural network using
algorithm backpropagation. To help the recognition, the image is segmented by mapping method.
The result of testing shows that for each type of character recognized was 81.74%
recognized for the succeed segmentation and 67.53% for the failed one. The recognition of other
type of character from the recognized one result is 55.47% for the succeed segmentation and 46.65%
for the failed one.
Keyword: Printed Characters Recognition, Backpropagation, Digital Images.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.