Image Retrieval And Categorization Based On Dictionary Of Block Visual Words

• Main Authors: Chen, Yuming, 陳育銘
• Other Authors: Yang, Naichung
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/05619489498817016246

碩士 === 義守大學 === 資訊工程學系 === 100 === The text-based retrieval systems query databases by keywords, they can not reflect human’s semantics. Currently, the content-based retrieval (CBIR) systems extract low-level features from images; however, the gap between high- and low-level semantics is still exists. To reduce this gap, a possible way is to extracted effective features form image. We first partition an image into 4×4 sub-images, then extract their image features. We propose to use the macro and micro features, which is consistent with human visualization, to increase the retrieval performance. After training, the extracted features are merged as visual words. Finally, the visual dictionary is composed of the collected visual words. Experimental results show that the proposed visual dictionary achieves good retrieval performance.