基于多特征融合和二维投影非负矩阵分解的图像检索

Image Retrieval Based on Feature Fusion and Two-dimensional Projection Nonnegative Matrix Decomposition

文章提出了一种基于非负矩阵分解(Nonnegative Matrix Factorization.NMF)和多特征融合的图像检索模型。通过提取图像的颜色和纹理特征,进行NMF分解,得到NMF的基矩阵和样本的系数矩阵。利用二维主成分分析(2DPCA)的思想对系数矩阵降维,然后通过特征加权的方法比较检索结果。文章使用500幅人物图像组成的图像库进行试验,该方法利用了图像的多个特征和2DPCA思想,使得文章中的方法提高了检索的查准率,而且检索速度优于非负矩阵分解和二维主成分分析。

This paper presents an image retrieval model based on non-negative matrix factorization（Nonnegative Matrix Factorization.NMF） and multi-feature fusion. By extracting color and texture features of images for NMF decomposition, the base matrix of NMF and the coefficient matrices of samples are obtained. The two-dimensional principal component analysis（2 DPCA） idea is used for dimension reduction of coefficient matrix, and then the retrieval results are compared by using the method of feature weighting. This article uses the image library containing 500 character image for test, which uses the multiple characteristics of the image and 2DPCA idea, improving the precision of retrieval in the article, and the retrieval speed is superior to the decomposition of nonnegative matrices and the two-dimensional principal component analysis.