蕾丝花边的改进型纹理特征检索方法  

Lace retrieval method based on improved texture feature

作　　者：曹霞[1] 李岳阳[1] 罗海驰 蒋高明[1] 丛洪莲[1] 

机构地区：[1]江南大学教育部针织技术工程研究中心,江苏无锡214122 [2]江南大学轻工过程先进控制教育部重点实验室,江苏无锡214122

出　　处：《纺织学报》2016年第6期142-147,154,共7页Journal of Textile Research

基　　金：中央高校基本科研业务费专项资金项目(JUSRP51404A;JUSRP211A38);江苏省产学研联合创新资金-前瞻性联合研究项目(BY2015019-11;BY2013015-14);江苏省高校优势学科建设工程资助项目(苏政办[2014]37号)

摘　　要：针对传统的蕾丝花边检索主要依赖于人的视觉检测及文本检索,存在信息不稳定、效率低、检索效果不可靠的现象,提出了一种基于层次匹配下多特征融合的蕾丝花边检索方法。通过运用图案纹理特征标识图像,首先分别用灰度共生矩阵、灰度梯度共生矩阵、局部二进制算子提取纹理特征进行匹配。然后将3种提取纹理特征方法,分别结合形状特征、不变矩特征量进行逐层匹配。最后将层次匹配下各个纹理特征进行融合,弥补了单个匹配方法的不足,同时在蕾丝花边库中验证所用检索方法的正确率。分析结果表明,这种方法优于任意单个的蕾丝花边匹配方法,能较好地实现蕾丝花边检索,有效地提高图案检索的可靠度和准确率。Conventional lace pattern retrieval mainly relies on the manual retrieval and text retrieval,while the text retrieval marks image by language description which leads to the unstable annotation information,low efficiency and unreliable retrieval results. This work marked image by texture pattern feature. A lace retrieval algorithm containing classification selection and classifier fusion by hierarchical matching was proposed,and it made up the deficiency of the single matching method. Three image-based methods,such as gray level co-occurrence matrix,gray level-gradient co-occurrence matrix and local binary pattern operator,were fused by means of geometry features and invariant moments for match-bylevel,respectively. Experimental results indicate that the performance of fusion-based method is better than any single method,and it can achieve the lace retrieval and improves the reliability and accuracy of image retrieval effectively.

关 键 词：图像检索 蕾丝花边 纹理特征匹配 层次匹配 特征融合 特征提取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]