基于双密度小波变换的纹理图像检索

为了进一步提高纹理图像的检索性能，提出了一种基于双密度小波变换算法．该算法根据双密度小波分解变换的特点，从系数角度出发首先对子带进行组合，然后提取一阶和二阶统计矩并将结果作为纹理的特征用于图像检索．由于组合双密度小波变换采用了过采样，具有时移不变性，所以据此生成的算法具有特征数少、检索精度高等特点．对比实验结果表明，该算法的检索精度比单小波和双密度小波变换分别提高了10％和7％，性能最好的是一阶和二阶统计矩组合的方法．

Texture Image Retrieval Based on Double Density Wavelet Transform

In order to enhance the performance of the texture image retrieval,a new algorithm based on the double density wavelet transform(DDWT) was presented,which was obtained by interleaving double density wavelet transforms(IDDWT) using the characteristics of the double density wavelet decomposition and computing the first-order and the second-order statistical parameters of IDDWT as the texture feature for image retrieval.In comparison with the traditional wavelet,the pyramid discrete wavelet decomposition transforms(PDWT) utilizes the oversampled framework and has time shift invariant.This algorithm is better than those of PDWT and DDWT under the same feature extraction method and the same similarity measure.In the contrast experiment, the result shows that the retrieval efficiency of this algorithm is increased by 10% and 7%,respectively,for image retrieval.The best performance is achieved with combinatorial method of the first and second order statistical moments.