Abstract: Focusing on the problem of low retrieval rate and retrieval speed in contourlet retrieval systems，a wavelet-contourlet transform based texture image retrieval system is proposed.In the system，contourlet transform is carried out for most detailed sub-bands.Each sub-band energy and standard deviations in transform domain are cascaded to form feature vectors，and the similarity metric is Canberra distance.Comparison is made on contourlet and wavelet-contourlet transform retrieval systems through an experiment on 109 texture images with 640×640 pixels size from Brodatz album.Experimental results show that the latter one has advantages on retrieval rates，retrieval speed and storage room for feature vectors.

Key words: wavelet-contourlet transform, contourlet transform, texture image, retrieval system, retrieval rate

摘要： 针对轮廓波检索系统检索率较低、检索速度较慢等问题，提出一种小波-轮廓波纹理图像检索系统。该系统在最细节尺度子空间采用轮廓波变换，而在其他子空间采用小波变换，以变换域子带系数的能量和标准偏差序列作为特征向量，Canberra距离为相似度标准。采用Brodatz纹理数据库中的109幅640×640的纹理图像为实验对象，比较了基于同样架构的轮廓波变换和小波-轮廓波纹理检索系统的性能。实验结果表明：后者具有较高的检索率、较快的检索速度和较少的特征向量数据库存储空间。

关键词: 小波-轮廓波变换, 轮廓波变换, 纹理图像, 检索系统, 检索率