无下采样轮廓波广义高斯纹理检索系统

Non-Subsampled Contourlet Generalized Gaussian Texture Retrieval System

针对轮廓波变换纹理图像检索系统检索率较低的问题,提出一种基于广义高斯分布和无下采样轮廓波变换的纹理图像检索系统.该系统采用的特征向量由无下采样轮廓波变换域各子带内系数广义高斯分布的尺度参数和形状参数连接而成,相似度度量标准使用K-L距离.实验结果表明:采用相同的系统结构和分解结构参数,在基本相同的检索时间内,无下采样轮廓波检索系统比基本轮廓波检索系统具有更高的检索率.

With focusing on the problem of low retrieval rate of texture image retrieval system based on contourlet transform,a texture image retrieval system based on generalized Gaussian distribution（GGD）and non-subsampled contourlet-transform is proposed.The feature vectors in the system is constructed by cascading scale and shape parameters of GGD for every sub-band in contourlet domain,and the similarity metric used here is K-L distance.Experimental results show that non-subsampled contourlet transform based image retrieval system is superior to that of the original contourlet transform under the same system structure and decomposition parameter with almost same retrieval time.