MADURA BATIK DETECTOR USING COMPUTER VISION-BASED ETHNOMATHEMATICAL APPROACH
Madura batik is one of the traditional cloth from Indonesia which has a variety of motifs. Madura batik cloth motifs can be identified by detection. The detection of batik cloth is done because there are still many people who do not know the diversity of Madura batik motifs. Detection of Madura batik cloth from various motifs can be done by humans but the results can be subjective and will take more time. So that the detection research is carried out using an ethnomatic approach based on Computer Vision. The research method for detecting Madura batik cloth is carried out in several stages, namely, data collection, system architecture, feature extraction, and artificial neural network. The batik data collected are 43 datasets originating from Tanjung Bumi and have motifs in the form of circles, rectangles, triangles, and others. System architecture is needed in detecting motifs on batik images so that the system that can be made functions optimally and as expected. The feature extraction stage is carried out to determine the identity of an image, feature extraction used in this study includes color features using the HSV moments method, shape features using the Hu-Moment method, and texture features using the GLCM method. Artificial neural network is done to classify the human brain oriented that can process information. The method contained in the Artificial Neural Network is Backpropagation, a multilayer network consisting of several layers, namely, input, hidden, and output layers. The results of this research are the detection of Madura batik motifs in the form of image processing, feature extraction in images, and system evaluation. Image processing in this study consists of grayscalling, histogram equalization, and canny processes. The results of the features of an image in this study have 29 features where 6 features are from color extraction, 7 features are from shape extraction and 16 features are from texture extraction. Evaluation of the system in the confusion matrix produces precision values = 100%, recall = 100%, and f-1 score = 100%.
batik; system architecture; feature extraction; artificial neural network