Fast Discrete Curvelet Transform And HSV Color Features For Batik Image Classification

Batik is one of the cultural heritages in Indonesia. Batik has many types spread around Indonesia. Related to the diversity of batik, an effort to develop a database to preserve batik information is required. Searching batik information from the database by using keywords such as the province name where a batik came from, sometimes is difficult. In some cases, people only has a batik image without knowing any additional information, such as motif name and it's origin. Attaching a modul to classify batik image automatically into the database will be very useful, so that people can search more information about batik by inputting a batik image. This research proposes batik image classification using Fast Discrete Curvelet Transform (FDCT) and Hue Saturation Value ( HSV) space as the representation of texture and color features, and K Nearest Neighbour ( KNN) as the classifier. The experiment give a good result, which is showed by the worst classification error rate 3.33% for combined features vector.