基于内容的图像检索中的一种多粒度组合查询方法

A Combined Method of Multiple Level Query Result for Content-based Image Re trieval

将图像分成不同层次度量时,由于不同层对最后结果所起的作用不同,因而在合并不同层产生的结果集时,应该考虑应用非线性变换。文章在图像信息的层次化描述模型的基础上,提出了一种用神经网络组合不同层的查询结果的方法。实验表明该方法是可行的。

In image similarity measure,the higher s imilarity is,the more similar the images are.However,as the function in each level,the final query result is not union or intersection of all levels'.Ther efore,a non-linear transformation should beintroduced.This paper proposed a c ombined method of multiple level query result by neural network based multiple l evel description model.Results show the availability of the proposed method.