Ekstraksi Fitur Perpotongan dan Lengkungan untuk Mengenali Huruf Cetak

Abstract

Human techniques in manipulating picture are more and more, especially in character recognition. Various types of techniques to manipulate it also has advantages and disadvantages. One technique that is often used is to use artificial neural network. While the recognition technique using the feature extraction is still rarely used.
Therefore, an application was made to recognize a feature extraction of capital letters in the image. The technique used is to find the intersection or edge and curvature features from images containing capital letters so that it can be done the recognition of features from such capital letters.
Through the test results, has obtained an example of the introduction of the intersection and curvature features stored in the matrix as compared to the basic structure of the letter. Matrix generated from this testing , for some particular font will be quite satisfactory result . But there are still shortcomings in recognizing letters in a font that is not symmetrical.