摘要
轮廓波变换纹理检索系统检索率比较低的根本原因在于轮廓波变换较低的移不变水平和变换域系数较高的振荡性。为了克服轮廓波变换的这些缺陷,提出了一种基于映射的复Contourlet-1.3轮廓波变换,并证明了该变换的移不变性质。在该变换的基础上,采用子带系数的能量和标准偏差序列作为特征向量,以Canberra距离为相似度度量标准,提出了一种纹理图像检索方法。实验结果表明:在特征向量长度、检索时间、所需存储空间相同的情况下,基于映射的复Contourlet-1.3检索系统比基于同样架构的实数轮廓波变换、实数Contourlet-2.3、基于映射的复轮廓波变换、基于映射的复Contourlet-2.3等检索系统有更高的检索率。
The ultimate reason of low retrieval rate of contourlet transform texture image retrieval system lies in the coefficients shift sensitive characters and high oscillating in transform domain. In order to overcome the defects of contourlet transform, a mapping-based contourlet-1.3 transform was proposed, and the shift invariant character was proved. A texture image retrieval system based on the new transform was proposed in which the feature vectors are formed by cascading the energy and standard deviation of each sub-band in the transform domain, and Canberra distance was used as similarity metric. Experimental results show that the image retrieval system is superior to that of the original contourlet transform, contourlet-2.3, mapping-based contourlet transform and mapping-based contourlet-2.3 transform with the same length of feature vectors, retrieval time and memory.
出处
《光电工程》
CAS
CSCD
北大核心
2010年第10期83-88,共6页
Opto-Electronic Engineering
基金
国家自然科学基金资助项目(60572048)
河南省教育厅自然科学研究计划项目(2010B120009)