融合高级语义特征的色纺织物图像快速检索

doi:10.13475/j.fzxb.20210310406

纺织学报 ›› 2022, Vol. 43 ›› Issue (03): 89-94.doi: 10.13475/j.fzxb.20210310406

融合高级语义特征的色纺织物图像快速检索

谷迁¹^,², 袁理¹^,²(), 杨亚莉¹, 刘军平³

1.武汉纺织大学电子与电气工程学院, 湖北武汉 430200
2.武汉纺织大学湖北省功能纤维加工及检测工程技术研究中心, 湖北武汉 430200
3.武汉纺织大学数学与计算机学院, 湖北武汉 430200

收稿日期:2021-03-20 修回日期:2021-11-21 出版日期:2022-03-15 发布日期:2022-03-29
通讯作者: 袁理
作者简介:谷迁(1996—),男,硕士生。主要研究方向为色纺织物的图像处理。
基金资助:
湖北省自然科学基金项目(2014CFB754);湖北省教育厅科学技术研究计划青年人才项目(Q20141607);中国纺织工业联合会科技指导性计划项目(2018035);中国纺织工业联合会科技指导性计划项目(2014072);湖北省教育厅科学研究计划指导性项目(B2021106)

Fast image retrieval of colored spun fabrics on high-level semantic features

GU Qian¹^,², YUAN Li¹^,²(), YANG Yali¹, LIU Junping³

1. School of Electronic and Electrical Engineering, Wuhan Textile University, Wuhan, Hubei 430200, China
2. Hubei Functional Fiber Processing and Testing Engineering Technology Research Center, Wuhan Textile University,Wuhan, Hubei 430200, China
3. School of Mathematics and Computer, Wuhan Textile University,Wuhan, Hubei 430200, China

Received:2021-03-20 Revised:2021-11-21 Published:2022-03-15 Online:2022-03-29
Contact: YUAN Li

摘要/Abstract

摘要：

为提高色纺织物图像检索系统的有效性与执行效率,建立了融合高级语义特征的色纺织物图像检索系统。该系统利用深度卷积神经网络对色纺织物图像的风格等高级语义特征进行提取,并融合图像的局部二值模式与方向梯度直方图等浅层纹理特征构建层次化检索系统。与此同时,采用二进制哈希编码对提取的高维特征向量进行降维。采用该系统对具有9种不同纹理风格的色纺织物样本图像进行检索,结果表明,该系统的Top-10查全率与平均准确率分别达到了97.37%和87.54%,具有理想的有效性与鲁棒性;相较于直接利用高级语义特征进行检索,提出的检索方法执行效率提升约750倍。

关键词: 色纺织物, 图像检索, 高级语义特征, 二进制哈希码, 层次化特征

Abstract:

In order to improve the effectiveness and execution efficiency of image retrieval systems for colored spun fabrics, an image retrieval system based on low-order features and high-level semantic features was established. The system uses the deep convolution neural network to extract the style semantic features of colored spun fabrics, and then constructs a hierarchical retrieval system by fusing low-order texture features such as local binary pattern and directional gradient histogram. Binary hash coding was used to reduce the dimension of the extracted high-dimensional feature vector. The results showed that for the retrieval of the 9 sample images of colored spun fabrics with different texture styles, the Top-10 recall and average accuracy of the system reached 97.37% and 87.54%, respectively, indicating good effectiveness and robustness. Compared with the direct use of advanced semantic features, the retrieval time of the proposed method is about 750 times faster.

Key words: colored spun fabric, image retrieval, high-level semantic feature, binary hash code, hierarchical characteristic

中图分类号:

TS101.9

谷迁, 袁理, 杨亚莉, 刘军平. 融合高级语义特征的色纺织物图像快速检索[J]. 纺织学报, 2022, 43(03): 89-94.

GU Qian, YUAN Li, YANG Yali, LIU Junping. Fast image retrieval of colored spun fabrics on high-level semantic features[J]. Journal of Textile Research, 2022, 43(03): 89-94.

图/表 12

图1

图2

表1

图3

表2

表3

表4

表5

图4

图5

图6

表6

参考文献 12

[1]	方珍红, 谢明元, 杨玲. 一种新的基于加权颜色特征的图像检索算法[J]. 激光与红外, 2013,43(4):471-474.
	FANG Zhenhong, XIE Mingyuan, YANG Ling. New retrieval algorithm based on weighted color feature[J]. Laser & Infrared, 2013,43(4):471-474.
[2]	崔红静, 景军锋, 张缓缓, 等. 融合颜色和边缘特征的织物图像检索算法[J]. 棉纺织技术, 2019(10):35-39.
	CUI Hongjing, JING Junfeng, ZHANG Huanhuan, et al. Fabric image retrieval method of blending the color and edge feature[J]. Cotton Textile Technology, 2019(10):35-39.
[3]	向忠, 何旋, 钱淼, 等. 基于边缘和颜色特征的织物印花花型检索[J]. 纺织学报, 2018,39(5):137-143.
	XIANG Zhong, HE Xuan, QIAN Miao, et al. Printed fabric pattern retrieval based on edge and color features[J]. Journal of Textile Research, 2018,39(5):137-143.
[4]	牛盼盼, 王向阳, 周璐, 等. 基于多语义特征的彩色图像检索技术研究[J]. 计算机科学, 2009,36(3):226-231.
	NIU Panpan, WANG Xiangyang, ZHOU Lu, et al. Content-based color image retrieval using multi-semantics[J]. Computer Science, 2009,36(3):226-231.
[5]	王志伟, 普园媛, 王鑫, 等. 基于多特征融合的多尺度服装图像精准化检索[J]. 计算机学报, 2020,43(4):740-754.
	WANG Zhiwei, PU Yuanyuan, WANG Xin, et al. Accurate retrieval of multi-scale clothing images based on multi-feature fusion[J]. Chinese Journal of Computers, 2020,43(4):740-754.
[6]	孙洁, 丁笑君, 杜磊, 等. 基于卷积神经网络的织物图像特征提取与检索研究进展[J]. 纺织学报, 2019,40(12):146-151.
	SUN Jie, DING Xiaojun, DU Lei, et al. Research progress of fabric image feature extraction and retrieval based on convolutional neural network[J]. Journal of Textile Research, 2019,40(12):146-151.
[7]	徐慧, 白美丽, 万韬阮, 等. 基于深度学习的服装图像语义分析与检索推荐[J]. 纺织高校基础科学学报, 2020,33(3):64-72.
	XU Hui, BAI Meili, WAN Taoruan, et al. Image semantic analysis and retrieval recommendation for clothing based on deep learning[J]. Basic Sciences Journal of Textile Universities, 2020,33(3):64-72.
[8]	刘沐黎, 袁理, 杨亚莉, 等. 基于混合色彩空间独立特征的色纺面料颜色表征模型[J]. 纺织学报, 2019,40(9):62-69.
	LIU Muli, YUAN Li, YANG Yali, et al. Color representation model for pre-colored fiber blends based on independent component features from mixed color space[J]. Journal of Textile Research, 2019,40(9):62-69.
[9]	袁理, 代乔民, 付顺林, 等. 基于原色纤维混配色织物的呈色特性与影响因素分析[J]. 纺织学报, 2018,39(10):38-43.
	YUAN Li, DAI Qiaomin, FU Shunlin, et al. Analysis of color rendering characteristics and influencing factors of primary color fiber mixed color fabrics[J]. Journal of Textile Research, 2018,39(10):38-43. doi: 10.1177/004051756903900107
[10]	OJALA T, PIETIK A M, MAENPAA T. Multiresolution grayscale and rotation invariant texture classiflcation with localbinary patterns[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002,24(7):971-987. doi: 10.1109/TPAMI.2002.1017623
[11]	赵树志, 狄岚, 何锐波. 基于改进判别性完整局部二值模式与格分割的织物瑕疵检测方法[J]. 纺织学报, 2018,39(9):57-64.
	ZHAO Shuzhi, DI Lan, HE Ruibo. Fabric defect inspection based on modified discriminant complete local binary pattern and lattice segmentation[J]. Journal of Textile Research, 2018,39(9):57-64.
[12]	夏海浜, 黄鸿云, 丁佐华. 基于迁移学习与支持向量机的服装舒适度评估[J]. 纺织学报, 2020,41(6):125-131.
	XIA Haibang, HUANG Hongyun, DING Zuohua. Clothing comfort evaluation based on transfer learning and support vector machine[J]. Journal of Textile Research, 2020,41(6):125-131.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

样本名称	线密度/tex	面密度/(g·m^-2)	样本数量
春芽纱	22.5	170	56
春芽竹节纱	18.2	160	170
幻影纱	19.5	160	64
迷你幻影纱	22.5	170	64
霓彩纱	19.5	160	60
赛络色纺纱	19.5	160	40
手纺纱	29.2	175	160
水蚊蝇带纱	36.5	190	10
星彩纱	18.2	140	60

哈希码长度	查全率/%	平均准确率/%
16	90.37	46.58
32	91.46	48.12
64	94.52	49.36
128	96.03	50.88
256	98.87	50.97

LBP算子参数		查全率/%	平均准确率/%
P	R	查全率/%	平均准确率/%
8	1	94.68	81.62
8	2	97.04	83.77
16	1	96.41	79.34
16	2	96.67	77.57

HOG算子参数		查全率/%	平均准确率/%
c	b	查全率/%	平均准确率/%
36	4	94.03	82.77
36	9	93.85	80.68
64	4	96.57	83.32
64	9	95.76	82.95

样本名称	查全率/%	平均准确率/%
春芽纱	97.55	87.32
春芽竹节纱	98.21	89.28
幻影纱	97.56	87.78
迷你幻影纱	97.85	88.14
霓彩纱	97.63	87.03
赛络色纺纱	96.11	86.95
手纺纱	98.37	89.47
水蚊蝇带纱	95.21	85.61
星彩纱	97.84	86.28

融合高级语义特征的色纺织物图像快速检索

Fast image retrieval of colored spun fabrics on high-level semantic features

RichHTML

PDF (PC)

摘要/Abstract

引用本文

使用本文

图/表 12

参考文献 12

相关文章 10

Metrics

本文评价

推荐阅读 0

实验方法	查全率/%	平均准确率/%
方法1	67.54	53.83
方法2	85.62	61.43
方法3	78.65	66.57
方法4	84.22	80.33
本文方法	97.37	87.54

[1]	龚雪, 袁理, 刘军平, 杨亚莉, 刘沐黎, 柯政涛, 鄢煜尘. 混合色彩空间与多核学习的色纺织物组织点识别[J]. 纺织学报, 2020, 41(05): 58-65.
[2]	孙洁, 丁笑君, 杜磊, 李秦曼, 邹奉元. 基于卷积神经网络的织物图像特征提取与检索研究进展[J]. 纺织学报, 2019, 40(12): 146-151.
[3]	刘沐黎, 袁理, 杨亚莉, 刘军平, 程哲, 龚雪, 纪昌俊, 鄢煜尘. 基于混合色彩空间独立特征的色纺面料颜色表征模型[J]. 纺织学报, 2019, 40(09): 62-69.
[4]	吴传彬, 刘骊, 付晓东, 刘利军, 黄青松. 结合显著区域检测和手绘草图的服装图像检索[J]. 纺织学报, 2019, 40(07): 174-181.
[5]	袁理, 王丹书, 谷迁, 屠劭杰, 熊莹, 袁浩然, 刘军平, 鄢煜尘. 基于光谱泛相似测度的色纺纱线与织物间呈色规律[J]. 纺织学报, 2019, 40(02): 30-37.
[6]	袁理代乔民付顺林郑力文鄢煜尘. 基于原色纤维混配色织物的呈色特性与影响因素分析[J]. 纺织学报, 2018, 39(10): 38-43.
[7]	向忠何旋钱淼胡旭东. 基于边缘和颜色特征的织物印花花型检索[J]. 纺织学报, 2018, 39(05): 137-143.
[8]	曹霞李岳阳罗海驰蒋高明丛洪莲. 蕾丝花边的改进型纹理特征检索方法[J]. 纺织学报, 2016, 37(06): 142-154.
[9]	林翔宇张华熊. 基于图像边缘特征的领带花型检索方法[J]. 纺织学报, 2015, 36(05): 123-126.
[10]	项明姚雪存崔振东华铨平. 利用相关反馈技术和形状特征的服装图像检索算法[J]. 纺织学报, 2014, 35(2): 94-0.