摘要
提出了一种基于小波变换的分级图像检索方法,对图像的色彩分量进行Daubechies小波变换,提取低频及高频信息构建特征向量.将查询图像与数据库中图像的特征向量进行方差、不变矩和向量夹角余弦值的分级筛选,取得了较好的检索效果.
This paper discussed a step by step image retrieval method using wavelet transform. The indexing algorithm applies a Daubechies＇wavelet transform for the color component. The wavelet coefficients in the lower and higher frequency are stored as feature vectors. Compare the similarities of feature vectors between the query and the images in the database using variance of feature vectors, invariant matrix and cosine value of the vectors. Step by step retrieval has preferable retrieval effect.
出处
《四川大学学报（自然科学版）》
CAS
CSCD
北大核心
2007年第2期329-333,共5页
Journal of Sichuan University(Natural Science Edition)
关键词
基于内容的检索
小波变换
图像数据库
content based retrieval, wavelet transform, image database
