基于DTCWT域统计特征融合的纹理图像检索  

Texture image retrieval based on statistical feature fusion in DTCWT domain

作　　者：曲怀敬[1] 王恒斌 徐佳 王纪委 魏亚南 QU Huaijing;WANG Hengbin;XU Jia;WANG Jiwei;WEI Yanan(School of Information and Electrical Engineering, Shandong Jianzhu University, Jinan 250101, China)

机构地区：[1]山东建筑大学信息与电气工程学院,山东济南250101

出　　处：《山东建筑大学学报》2020年第3期28-35,共8页Journal of Shandong Jianzhu University

基　　金：山东省自然科学基金项目(ZR2014FM016);山东省重大科技创新工程项目(2019JZZY010120)。

摘　　要：在多尺度变换域,将各子带系数的统计特征进行互补融合可以有效地提高纹理图像检索的性能。文章利用双树复小波变换提出一种新的将低频子带系数的能量特征、高频子带幅值系数的Weibull分布参数特征以及相对相位系数的wrapped Cauchy分布参数特征相融合的纹理图像检索方法,采用VisTex纹理图像库进行检索。结果表明:采用多类系数统计特征的互补融合,以及最优的相似性测度加权组合,能够显著地提高纹理图像检索系统的平均检索率;与现有的7种纹理图像检索方法相比较,所获得的较高平均检索率为86.74%。In the multi-scale transform domain,the performance of texture image retrieval can be effectively improved by the complementary fusion of the statistical features of each sub-band coefficients.This paper proposes a new texture image retrieval method which combines the energy feature of the low-frequency subband coefficients,the Weibull distribution parameter feature of the high-frequency subband amplitude coefficients and the wrapped Cauchy distribution parameter feature of the relative phase coefficients by using the double-tree complex wavelet transform.The experimental results show that the average retrieval rate of texture image retrieval system can be significantly improved by using the complementary fusion of multi-class coefficients statistical features and the optimal weighted combination of similarity measures.Compared with the existing 7 texture image retrieval methods,the proposed approach achieves a higher average retrieval rate with 86.74%.

关 键 词：双树复小波变换 纹理图像检索 统计特征 特征融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]