纺织学报 ›› 2019, Vol. 40 ›› Issue (06): 117-124.doi: 10.13475/j.fzxb.20180704708

• 管理与信息化 • 上一篇    下一篇

多纹理分级融合的织物缺陷检测算法

朱浩1, 丁辉1,2,3, 尚媛园1,2,3, 邵珠宏1,2,4   

  1. 1.首都师范大学 信息工程学院, 北京 100048
    2.高可靠嵌入式系统技术北京市工程技术研究中心,北京 100048
    3.北京成像理论与技术高精尖创新中心, 北京 100048
    4.北京数学与信息交叉科学协同创新中心, 北京 100048
  • 收稿日期:2018-07-19 修回日期:2019-02-26 出版日期:2019-06-15 发布日期:2019-06-25
  • 通讯作者: 丁辉
  • 基金资助:
    国家自然科学基金项目(61876112, 61303104, 61601311, 61603022);北京市属高校高水平教师队伍建设支持计划项目(CIT&TCD20170322);北京市教委科研计划项目(SQKM201810028018);首都师范大学创新团队项目(PXM19530050151)

Defect detection algorithm for multiple texture hierarchical fusion fabric

ZHU Hao1, DING Hui1,2,3, SHANG Yuanyuan1,2,3, SHAO Zhuhong1,2,4   

  1. 1. College of Information Engineering, Capital Normal University, Beijing 100048, China
    2. Beijing Engineering Research Center of High Reliable Embedded System, Beijing 100048, China
    3. Beijing Advanced Innovation Center for Imaging Theory and Technology, Beijing 100048, China
    4. Collaborative Innovation Center for Mathematics and Information of Beijing, Beijing 100048, China
  • Received:2018-07-19 Revised:2019-02-26 Online:2019-06-15 Published:2019-06-25
  • Contact: DING Hui

摘要:

针对织物缺陷检测过程中纹理分布的复杂多样性引起误检和漏检的问题,结合织物纹理周期性特点,提出一种多纹理分级融合的织物缺陷检测算法。在检测过程中,首先利用织物缺陷图像的Tamura粗糙度图,对缺陷区域进行初步定位和自适应性生长,将初步定位的区域映射到原始织物图像中;其次根据织物图像的周期性分布特征,对初步定位区域进行分块,提取图像块的局部相位量化(LPQ)特征、Tamura特征,并将2种特征融合;然后计算融合特征与正常块特征的相似度,获取相似度图;最后将初步定位区域的经纬向特征图与相似度特征图融合,检测缺陷存在的区域。经TILDA织物纹理库数据的实验测试结果表明,缺陷区域的初步定位和自适应生长,降低了缺陷检测过程的冗余度,提高了检测效率,避免了织物缺陷检测过程中的误检和漏检情况。

关键词: 织物缺陷检测, 织物纹理, 特征融合, Tamura特征, 局部相位量化特征

Abstract: Aim

ing at the false detection and missing detection caused by complexity and diversity of texture distribution in fabric defect detection, considering the periodicity of fabric texture, an algorithm of multi-texture gradation fusion for fabric defect detection was proposed. In the process of testing, firstly, the defect region was subjected to primary positioning and self-adaptive growth by using the Tamura roughness graph of the fabric defect image, then the primary positioned region was mapped to the original fabric image. The primary positioned region was blocked and the local phase quantization (LPQ) texture feature and Tamura texture feature of each image block were extracted, and the two different texture features were fused. The similarity between the fusion feature and the normal block feature was calculated to obtain the similarity image. Finally, the longitude and latitude feature map and the similarity feature map were fused to find the region of the defects in fabric images. The experimental results on TILDA dataset show that the new approach can reduce the redundancy of defect detection and improve the detection efficiency, and can avoid errors and omissions in the process of defect detection.

Key words: defect detection, fabric texture, feature fusion, Tamura feature, local phase quantization feature

中图分类号: 

  • TP399

图1

不同窗口大小的粗糙度图"

图2

不同纹理、周期分布织物图像"

图3

织物缺陷区域初步定位和自适应生长框图"

图4

缺陷识别结构图"

图5

缺陷初步定位和自适应生长结果"

图6

织物缺陷图检测结果 注:图中2~7列分别对应同一行中织物原图的检测结果。"

图7

实验对比图"

表1

不同纹理缺陷检测平均耗时"

织物编号 LBP LPQ 经纬向 Tamura LBP+经纬向 本文
试样1# 14.96 30.97 6.69 7.91 19.81 13.65
试样2# 14.73 31.12 6.72 7.89 19.93 13.15
试样3# 14.60 31.04 6.71 7.92 19.94 12.95
试样4# 15.10 31.26 6.72 7.90 19.74 13.28

表2

不同纹理缺陷误检率和检出率"

织物
编号
LBP算法 LPQ算法 经纬向算法 LBP+经纬向算法 本文算法
检出率 误检率 检出率 误检率 检出率 误检率 检出率 误检率 检出率 误检率
试样1# 84.72 15.27 85.64 14.35 81.94 18.05 91.66 8.33 93.05 6.94
试样2# 83.33 16.66 83.79 16.20 80.55 19.44 94.44 5.55 93.05 6.94
试样3# 61.11 38.88 63.88 36.11 43.05 56.94 58.33 41.66 65.27 34.72
试样4# 54.16 45.83 56.01 43.98 41.66 58.33 69.44 30.55 72.77 27.77
[1] SRINIVASAN K, DASTOOR P H, RADHAKRISHNAIAH P, et al. FDAS: A knowledge-based framework for analysis of defects in woven textile structures[J]. Journal of the Textile Institute, 1992,83(3):431-448.
doi: 10.1080/00405009208631217
[2] LIU Bing, ZHANG Hong. Image retrieval algorithm based on convolutional neural network and manifold ranking[J]. Journal of Computer Applications, 2016,36(2):531-534.
[3] CHAN C H, PANG G K H. Fabric defect detection by Fourier analysis[J]. IEEE Transactions on Industry Applications, 2000,36(5):1267-1276.
doi: 10.1109/28.871274
[4] MAK K L, PENG P, YIU K F C. Fabric defect detection using morphological filters[J]. Image & Vision Computing, 2009,27(10):1585-1592.
[5] CHETVERIKOV D, HANBURY A. Finding defects in texture using regularity and local orientation[J]. Pattern Recognition, 2002,35(10):2165-2180.
doi: 10.1016/S0031-3203(01)00188-1
[6] ALATA O, RAMANANJARASOA C. Unsupervised textured image segmentation using 2-D quarter plane autoregressive model with four prediction supports[J]. Pattern Recognition Letters, 2005,26(8):1069-1081.
doi: 10.1016/j.patrec.2004.10.002
[7] 杜磊, 李立轻, 汪军, 等. 几种基于图像自适应阈值分割的织物疵点检测方法比较[J]. 纺织学报, 2014,35(6):56-60.
DU Lei, LI Liqing, WANG Jun, et al. Comparison of several fabric defect detection methods based on image adaptive threshold segmentation[J]. Journal of Textile Research, 2014,35(6):56-60.
[8] NGNA H Y T, PANG G K H, YUNG S P, et al. Wavelet based methods on patterned fabric defect detection[J]. Pattern Recognition, 2005,38(4):559-576.
doi: 10.1016/j.patcog.2004.07.009
[9] VERMAAK H, NSENGIYUMVAL P, LUWES N. Using the dual-tree complex wavelet transform for improved fabric defect detection[J]. Journal of Sensors, 2016,2016(7):1-8.
[10] BISSI L, BARUFF G, PLACIDIP , et al. Automated defect detection in uniform and structured fabrics using Gabor filters and PCA[J]. Journal of Visual Communication and Image Representation, 2013,24(7):838-845.
doi: 10.1016/j.jvcir.2013.05.011
[11] JING J, FAN X, LI P. Automated fabric defect detection based on multiple gabor filters and KPCA[J]. International Journal of Multimedia & Ubiquitous Engineering, 2016,11(6):93-106.
[12] JING J, LIU S, LI P, et al. The fabric defect detection based on CIE L*a*b* color space using 2-D Gabor filter[J]. Journal of the Textile Institute Proceedings & Abstracts, 2016,107(10):1305-1313.
[13] 杨晓波. 基于GMRF模型的统计特征畸变织物疵点识别[J]. 纺织学报, 2013,34(4):137-142.
YANG Xiaobo. Fabric defect detection of statistic aberration feature based on GMRF model[J]. Journal of Textile Research, 2013,34(4):137-142.
[14] TAMURA H, MORI S, YAMAWAKI T. Textural features corresponding to visual perception[J]. IEEE Transactions on Systems Man and Cybernetics, 1978,8(6):460-473.
doi: 10.1109/TSMC.1978.4309999
[15] SACHS J, GAO L. Item-level and subscale-level factoring of Biggs' learning process questionnaire (LPQ) in a mainland Chinese sample[J]. British Journal of Educational Psychology, 2011,70(3):405-418.
doi: 10.1348/000709900158209
[16] CHIKKERUR S, GOVINDARAJU V, CARTWRIG A N. Fingerprint image enhancement using STFT analy-sis[J]. Pattern Recognition, 2007,40(1):198-211.
doi: 10.1016/j.patcog.2006.05.036
[17] 周建, 王静安, 高卫东. 应用局部纹理特征的无监督织物瑕疵检测算法[J]. 纺织学报, 2016,37(12):43-48.
ZHOU Jian, WANG Jingan, GAO Weidong. An unsupervised fabric defect Detection algorithm based on Local texture feature[J]. Journal of Textile Research, 2016,37(12):43-48.
[18] 王钢, 周建, 汪军, 等. 基于奇异值分解的机织物瑕疵检测算法[J]. 纺织学报, 2014,35(7):61-66.
WANG Gang, ZHOU Jian, WANG Jun, et al. Flaw detection algorithm for woven fabrics based on singular value decomposition[J]. Journal of Textile Research, 2014,35(7):61-66.
[1] 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(06): 125-131.
[2] 吴莹, 占竹, 汪军. 基于通用字典的局部机织物纹理稳定表征[J]. 纺织学报, 2019, 40(11): 45-49.
[3] 邢文宇, 邓娜, 辛斌杰, 于晨. 基于多特征融合图像分析技术的羊毛与羊绒鉴别[J]. 纺织学报, 2019, 40(03): 146-152.
[4] 吴莹 汪军. 基于K-SVD学习字典的机织物纹理表征及应用[J]. 纺织学报, 2018, 39(02): 165-170.
[5] 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.
[6] 薛乐 李立轻 汪军. 应用频谱图的机织物纹理分析[J]. 纺织学报, 2015, 36(07): 55-60.
[7] 汪秀琛, 李晓久. 基于复合条件的织物纹理模糊识别[J]. 纺织学报, 2012, 33(8): 55-58.
[8] 刘哲;李晓久. 织物纹理均匀度的计算机评价[J]. 纺织学报, 2009, 30(12): 125-128.
[9] 徐琪. 一种支持快速响应的服装供应链补货策略[J]. 纺织学报, 2009, 30(12): 129-133.
[10] 诸葛振荣;徐利祥. 利用纹理几何约束进行织物场景模拟[J]. 纺织学报, 2007, 28(6): 55-58.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!