Image Retrieval Using Dominant Color Distribution

• Main Authors: Wen-Cheng Wang, 王文政
• Other Authors: Chaur-Heh Hsieh
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/69975260773021799536

碩士 === 義守大學 === 資訊工程學系 === 92 === Recently, content-based image retrieval (CBIR) plays a very important role in large-scale multimedia systems. The CBIR employs image features such as color, texture and shape, to search the matched images from database. The human eyes are very sensitive to color difference, so color feature has been widely used in CBIR. This thesis studies an image retrieval system based on dominant colors defined by MPEG-7. To further improve the retrieving performance, we extract spatial distributions of the dominant colors. Based on the dominant colors and spatial distributions, two similarity metrics are designed, which are used for image matching. The experimental result shows that the new methods achieve better performance than the dominant color scheme in MPEG-7.