Identifikasi Isen-Isen Batik Menggunakan Backpropagation Dan Alihragam Wavelet

Abstract

Indonesian state has a valuable cultural heritage art form of batik. Batik has received recognition from the United Nations through UNESCO as a heritage of Indonesia since October 2, 2009. Motif consists of the main components form the principal ornament and isen. Isen generally has a simple shape and has a relatively small size. Isen used to embellish the principal ornament that can add charm, but batik lovers in general do not have enough knowledge on this cultural heritage. Isen batik samples used in this study include cecek pitu, hereangan, sisik melik, ukel, sisik, cecek, and galaran. Batik lovers in general less know the name isen batik cloth they wear. It is very unfortunate when batik lovers just wearing batik cloth without knowing the deeper will be the uniqueness of this batik. The lack of information about batik especially isen be one factor identification system built isen batik. The system will be built by using Artificial Neural Networks Backpropagation and wavelet transformation. The system will recognize the pattern of Isen-sen each to be taken characteristics and is then ready to be identified. The target to be achieved in this research is to provide accurate information about batik decoration isen especially in the detection of appropriate naming. The lack of information about isen batik has become one of the factors their ideas in the form of identification isen batik using Backpropagation Neural Network and Wavelet transformation. The system is designed to search for understanding patterns of isen as an object, and then after the obtained characteristics of the pattern, then the system will identify isen accordance with related names. This research involves two main parts: training and testing. Training begins by scanning the image of isen manually using a scanner to produce digital images. Digital image subjected to the initial process is preprocessing which includes, threshold, wavelet transformation, normalization, and after that the data is ready to be trained using back propagation neural network. In the form of weight training results will be stored in a data store as a trained image. The testing phase also involves preprocessing, as well as training and after normalization, the test images will be compared with the data already stored in the training data store. These test results in the form of ID is correlated with the name of isen related. It is hoped the system provides optimal accuracy and can be one source of information for batik in particular in identifying the name Isen Isen.