摘要
在研究与分析了形状特征提取算法和相似性度量算法的基础上,采用欧式距离法,利用UIUC图库中的鸟类、蝴蝶类和Caltech101图像库中chair类、strawberry类等4类图库,对Hu不变矩特征提取的图像检索算法进行验证,并对不同测试图库所得到的实验结果进行比较.从实验结果可以看出,对背景相对简单、对象的形态各异的图像还是具有较好的检索性能,可以得到较好的检索效果.
The paper has research and analyzed the arithmetic of shape features extraction and similarity metric,and adopted the Euclid distance metric,and according to the image database of the bird which from UIUC and the database of chair,strawberry from Caltech 101,the Hu invariants moments features extraction are validated.Compared these experiment results,at the different database,some image which has simple background and different shape object can be obtained the better retrieval effect.
出处
《河南科技学院学报》
2010年第1期93-96,共4页
Journal of Henan Institute of Science and Technology(Natural Science Edition)
基金
河南省教育厅"十一五"发展规划课题(2008-jkghagh-286)
关键词
图像检索
形状特征
相似性度量
图像数据库
HU不变矩
image retrieval
features extraction
similarity metric
image database
Hu invariants moments
