摘要
介绍了基于内容图像检索中多种颜色的特征提取和匹配算法,以及加权欧几里得距离、中心距的加权距离、直方图交集算法等。结合西北大学可视化研究所开发的数字博物馆影像数据库系统,对这些算法进行了实验比较。结果表明,加权距比欧氏距有所改善;而中心距算法简单,检索速度快;直方图交集法相对于加权欧氏距并没有明显改善。在实际应用中可先使用一种速度快的算法,再在查询结果上进行二次查询。
Introduces color feature extraction and matching algorithms in content-based image retrieval,such as weighted Euclidean-distance,weighted centre distance,histogram intersection algorithm,etc.Some algorithms were applied to a practicl digital museum image database information system developed by Visualization Institute of Northwest University, and presents a comparative experiment results.The experimentsal result shows that weighted Euclidean-distance is better than Euclidean-distance,and weighted center distance is not only simple but also high-speed. Histogram intersection algorithm has no apparent advantage over weighted Euclidean-distance.In praticality,a high-speed algorithm can be used firstly,secondly another algorithm can be used on the retrieval result.
出处
《微机发展》
2004年第2期86-88,共3页
Microcomputer Development
