| 基于可视特征度量距离的图像检索 |
| |
| 引用本文： | 刘传才. 基于可视特征度量距离的图像检索[J]. 福州大学学报(自然科学版), 2001, 29(2): 12-15 |
| |
| 作者姓名： | 刘传才 |
| |
| 作者单位： | 福州大学计算机科学与技术系, |
| |
| 基金项目： | 福建省自然科学基金!资助项目 (F0 0 0 13)，福建省教育厅科研基金!资助项目 (JS99- 0 4) |
| |
| 摘 要： | 研究预滤波、二进制集的划界、降维和查询优化的距离计算 ,并在 40 0 0幅测试图像集上 ,采用实验方法对高维直方图的索引策略与优化查询的距离计算进行了评估 .研究表明 ,非二次形式的距离度量与二次形式的距离度量一样有良好的性能 .此外 ,二进制集是直方图的一种有效而简捷的替代方法 ,其检索性能足以满足第一遍查询 .
|
| 关 键 词： | 可视特征 图像检索 颜色距离 纹理距离 |
| 文章编号： | 1000-2243(2001)02-0012-04 |
| 修稿时间： | 2000-09-26 |
Image indexing based on distance metrics of visual features |
| |
| LIU Chuan-cai. Image indexing based on distance metrics of visual features[J]. Journal of Fuzhou University(Natural Science Edition), 2001, 29(2): 12-15 |
| |
| Authors: | LIU Chuan-cai |
| |
| Affiliation: | LIU Chuan-cai (Department of Computer Science and Technology,Fuzhou University, Fuzhou,Fujian 350002,China) |
| |
| Abstract: | Pre-filtering, binary set bounding,dimensionality reduction and query optimized distance computation are investigated,and the retrieval strategy for high dimensional histograms and the query optimized distance computation by experiments on the test set of 4?000 images are evaluated. The research showed that the non-quadratic form distance metrics perform as well as the quadratic form metrics. Furthermore,binary sets are an efficient and compact alternative to histograms, and their performance suffices as a first pass query. |
| |
| Keywords: | visual feature image indexing color distance texture distance |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《福州大学学报(自然科学版)》浏览原始摘要信息 |
|
点击此处可从《福州大学学报(自然科学版)》下载全文 |
