摘要： 提出一种Contourlet变换域中基于矩特征的图像检索方法。为了提高方法对于噪声的鲁棒性，对图像进行Contourlet变换后，对变换结果进行阈值滤波，计算各子带系数的一阶几何矩作为图像的特征向量，采用欧氏距离进行相似度度量。该方法提取的图像特征能反映子带系数的空间分布特征，无需任何前提假设。实验结果表明，该方法具有较高的查准率，对噪声的容忍度高于现有算法。

关键词: 图像检索, Contourlet变换, 矩, 特征提取, 相似度度量

Abstract: This paper proposes an image retrieval method based on moment features in Contourlet transform domain. In order to improve its degree of noise tolerance, threshold filtering is applied in Contourlet transform results of the image, and the geometry moments of Contourlet coefficient with different scales and directions are obtained as the features of image. Similarity measurement is implemented by using Euclidean distance between the features of query image and that of each image in the image database. Because the method makes use of the spatial distribution features of Contourlet coefficients of images, it does not need any hypothesis. Experimental results show the superiority of the method in terms of the average rate of retrieving relevant images, and it is more robust than other algorithms.

Key words: image retrieval, Contourlet transform, moment, feature extraction, similarity measurement

中图分类号: 

• TN911.73