摘要
在分析以往基于边缘的图像检索算法的基础上,提出了一种基于多尺度欧拉矢量的图像检索算法。该算法针对灰度图像,以小波变换为基础,利用小波模极大值对图像进行多级小波分解,得到多尺度下的边缘图像,针对每一幅边缘图像,计算其欧拉数,构造一个欧拉矢量,作为对图像特征的描述。同时,考虑到欧拉矢量之间的相关性,设计了相关权值矩阵,提出采用马氏距离进行相似性度量。该矢量不仅能很好的表示图像的形状特征,而且具有平移、尺度和旋转不变性,同时它又是图像的一个整体特征,能刻画图像的拓扑结构。实验结果表明该算法计算简单,有效,匹配快速,检索结果比较理想。
A novel image retrieval based on Multi-scale Euler Vector is proposed by analyzing the existing retrieval methods on image edge. Firstly, the grey scale image is transformed by wavelet modulus maximum to get multi-scale edge images. Then the Euler number of each edge images is computed to extract the features of image. Consequently, each image is characterized by a multi-scale Euler vector in feature space. The co-relation between the features is taken into account, the co-relation matrix is constructed. Similarity is given by Mahalanobis Distance between two images＇ feature vectors. Multi-scale Euler Vector not only can capture the shape and spatial information of image but also can be invariant with respect to translation, scale and rotation of objects. At the same time, it can describe the topologic structure of image. The results from two image databases show that the retrieval performance is better and robust.
出处
《红外技术》
CSCD
北大核心
2006年第12期704-708,共5页
Infrared Technology
关键词
基于内容的图像检索
小波模极大值
欧拉数
content-based image retrieval
wavelet modulus maximum
Euler number
