A Framework for Image Data Management

• Main Authors: Chi-Hung Tsou, 鄒濟鴻
• Other Authors: Jia-Lien Hsu
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/10576522237028334591

碩士 === 輔仁大學 === 資訊工程學系 === 99 === When dealing with an increasing and huge amount of digital images, the traditional retrieval architecture associated with the keyword-based searching capability is not satisfying anymore, especially for non-experienced and naïve users. In this paper, we propose a framework for image data management, which supports content-based image retrieval. Also, we apply MPEG-7, one of most-popular multimedia content description interface, as the metadata specification of our framework. As preprocessing and extracting features from images, we choose angular radial transform, curvature scale space, edge histogram as features of images to be compared with user queries. In addition to MPEG-7 of sharing metadata, our framework is also followed with open architecture, in which more feature extraction modules and Ds/DS could be easily plugged into. By our experiment and performance evaluation, we show that the efficiency and effectiveness of our framework will be achieved.