基于Contourlet变换和Hu不变矩的图像检索算法

Image retrieval algorithm based on Contourlet transform and Hu invariant moments

文章提出一种基于Contourlet变换和Hu不变矩的图像检索算法。首先,对每幅图像进行Contourlet变换,得到低频子带与高频方向子带,把计算得到的低频子带的Hu不变矩和各个高频方向子带的均值与标准差作为图像的特征向量,利用Manhattan距离进行相似度度量,完成基于内容的图像检索。为对该文提出的算法的检索效果进行检验,分别与基于Contourlet变换特征的检索算法和基于Hu不变矩特征的检索算法等方法进行了对比实验研究。结果表明,该算法有效地融合了图像的纹理特征与低频子带的形状特征,较好地实现了基于内容的图像检索,平均查准率达到73.94%。

An image retrieval algorithm was proposed based on Contourlet transform (CT) and Hu invariant moments in this paper. Firstly, each image was decomposed into low frequency sub-band and high frequency sub-bands by using Contourlet transform. The Hu invariant moments of the low frequency sub-band coefficient, as well as the mean and the standard deviation of each high frequency sub-band coefficients were computed and used as image feature vector. Secondly, Manhattan distance was used as similarity measure between the query image and every image in the image database. After these two procedures, the content-based image retrieval was achieved. In order to evaluate the effect of the proposed algorithm, the algorithm based on CT and Hu invariant moments were tested respectively. Comparing the results of the average retrieval rate, the experimental results of the proposed algorithm were superior to other image retrieval algorithms. The proposed algorithm gets a higher average retrieval rate and the average retrieval rate is up to 73.94%.