Using multi-descriptors for khon image retrieval

• Main Authors: Jennisa Areeyapinan, Pizzanu Kanongchaiyos, Aram Kawewong
• Format: Conference Proceeding
• Published: 2018
• Subjects: Computer Science
• Online Access: https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=84893338812&origin=inward; http://cmuir.cmu.ac.th/jspui/handle/6653943832/52458
• Institution: Chiang Mai University

We present a method for Khon image retrieval using multi-descriptors. Khon is an ancient Thai cultural heritage that is very well-known from its gorgeous costumes and dance. Khon image retrieval can be adopted in various fields of work to preserve Thai culture and tradition. However, it is not trivial because of its complex and duplicated pattern caused by unique Thai line art. Thus, we integrate a Scale-invariant feature transform (SIFT) and Critical Point Filters (CPFs) to achieve accurate and fast Khon image retrieval. SIFT is used for details image such as Khon image. In order to reduce the time complexity for extracting key points using SIFT, we apply CPF which filter only the critical pixel of the image. From the experiment, our method can reduce computation time by 43.3% from SIFT and nearly 100% from CPF. Moreover, our method is preserve efficiency. © 2013 IEEE.