EKTRAKSI FITUR ANGKA JAWA MENGGUNAKAN DIAGONAL DISTANCE FEATURE DAN LONGEST RUN FEATURE

Abstract

Java script must always be preserved so that Indonesia has not lost its cultural value. Java script writing in the modern age can be done using a variety of media including computer, tablet, smartphone up. Results Java Script writing using media in the form of images. As a result, Java script writing in the picture unreadable text that can be recognized as a computer. One method of pattern recognition is a multi-layer perceptron (MLP). In this study, input MLP are 40 features in extraction using two methods: the diagonal distance feature and the longest run feature. Diagonal Distance feature extraction is extraction method that calculates the distance between the diagonal angle in temukannya image to black pixels of the characters in the image. While the longest run feature extraction is extraction method that calculates the number of black pixels along longest consecutive along the rows, columns, and two diagonal of the sub-image area. Samples training and testing samples of each amount to 25 sets. The success of the introduction of MLP using 40 features of 66.8% on training samples and 62.8% in the testing sample. The success of the introduction of MLP using 36 features of extraction longest run 100% on training samples and 94.4% in the testing sample. The success of the introduction of MLP uses 4 feature of extraction diagonal distance of 10% on training samples and testing samples.