Image Processing Technology for Motif Recognition Mandar Silk Fabric Android Based

Abstract

People use traditional equipment to produce silk fabrics for generations. Various motifs have been created and along with their development, sarong motifs are increasingly diverse and almost similar to one another. To be able to distinguish objectively, a technological approach will be used. An Image Processing technology that can identify the motifs and patterns of each silk fabric, this technology will be implemented so that it is used to help identify the names of each silk fabric motif or pattern, in addition, this technology will contribute and become a new way of preserving culture and customs. The bounding box technique is used to identify the motif pattern, then the Receiver Operating Character (ROC) method is used for the accuracy of the detection results. Research data in the form of 13 samples of sarong motifs will be collected and inventoried to create a training data table, after that, the user interface design of the application system will be built based on Android so that it is easy to use anytime. To ensure the consistency of the detection results, the system calibration is carried out with lighting techniques and identification distances. The results show that the precision value is 100% and the accuracy is 50%. However, we note that the recall value is only 60%. This shows that this system can only detect a small amount of saqbe sarong and there are still many sarongs that cannot be detected.