基于灰度梯度共生矩阵的快速布匹瑕疵检测算法

doi:10.16180/j.cnki.issn1007-7820.2025.07.008

电子科技 ›› 2025, Vol. 38 ›› Issue (7): 58-65.doi: 10.16180/j.cnki.issn1007-7820.2025.07.008

基于灰度梯度共生矩阵的快速布匹瑕疵检测算法

叶瑞帆¹, 刘瑜¹(), 沈杰¹, 任佳¹, 章小祥²

1.浙江理工大学信息科学与工程学院,浙江杭州 310018
2.浙江灿宇纺织有限公司,浙江衢州 324209

收稿日期:2024-01-08 修回日期:2024-01-25 出版日期:2025-07-15 发布日期:2025-07-10
通讯作者: 刘瑜(1975-),男,E-mail: liuyu@zstu.edu.cn,博士,副教授。研究方向:智能控制、机器人控制技术等。
作者简介:叶瑞帆(1998-),男,硕士研究生。研究方向:智能检测与控制。
沈杰(2000-),男,硕士研究生。研究方向:智能检测与控制。
基金资助:
浙江省“尖兵”研发攻关计划(2023C01062)

A Fast Fabric Defect Detection Algorithm Based on Gray Gradient Co-Occurrence Matrix

YE Ruifan¹, LIU Yu¹(), SHEN Jie¹, REN Jia¹, ZHANG Xiaoxiang²

1. School of Information Science and Engineering,Zhejiang Sci-Tech University,Hangzhou 310018,China
2. Zhejiang Canyu Textile Co., Ltd.,Quzhou 324209,China

Received:2024-01-08 Revised:2024-01-25 Online:2025-07-15 Published:2025-07-10
Supported by:
"Jianbing" Research and Development Plan of Zhejiang(2023C01062)

摘要/Abstract

摘要：

针对织物质量控制的瑕疵检测方法存在模型复杂、检测速度慢等问题,文中提出一种基于灰度梯度共生矩阵(Gray-Gradient Co-Occurrence Matrix, GGCM)的快速布匹瑕疵检测算法。在传统灰度共生矩阵(Gray Level Co-occurrence Matrix, GLCM)的基础上增添了对图像梯度信息的特征提取过程,并结合支持向量机(Support Vector Machine, SVM)对织物图像进行快速准确的检测分类。分析对比了GLCM与GGCM所提取的特征值,并使用SVM分类器检测布匹瑕疵。通过基于某纺织企业现场收集的织物图像数据集进行训练分类实验,结果表明加入梯度信息后检测效果显著提升,准确率达94.8%,精确率达93.9%。所提算法检测迅速,在提取特征后,每张图片检测仅耗时0.5 ms,适用于工业生产现场。

关键词: 布匹瑕疵检测, 纹理, 灰度梯度共生矩阵, 灰度共生矩阵, 特征值, 自适应中值滤波, 直方图均衡化, 支持向量机

Abstract:

In view of the problems of complex model and slow detection in fabric quality control, a fast fabric defect detection algorithm based on GGCM (Gray-Gradient Co-Occurrence Matrix) is proposed. Based on the traditional GLCM (Gray Level Co-Occurrence Matrix), this algorithm adds feature extraction of image gradient information, and combines with SVM (Support Vector Machine) to detect and classify fabric images quickly and accurately. The eigenvalues extracted from GLCM and GGCM are analyzed and compared, and the fabric defects are detected by SVM classifier. Through the training classification experiment based on the fabric image data set collected from the field of a textile enterprise, the results show that the detection effect is significantly improved after adding gradient information, the accuracy rate is 94.8%, and the accuracy rate is 93.9%. The algorithm is fast for detection, after extracting features, each image detection only takes 0.5 ms, which is suitable for industrial production sites.

Key words: fabric defect detection, texture, gray gradient co-occurrence matrix, gray level co-occurrence matrix, characteristic value, adaptive median filtering, histogram equalization, support vector machine

中图分类号:

TP181

叶瑞帆, 刘瑜, 沈杰, 任佳, 章小祥. 基于灰度梯度共生矩阵的快速布匹瑕疵检测算法[J]. 电子科技, 2025, 38(7): 58-65.

YE Ruifan, LIU Yu, SHEN Jie, REN Jia, ZHANG Xiaoxiang. A Fast Fabric Defect Detection Algorithm Based on Gray Gradient Co-Occurrence Matrix[J]. Electronic Science and Technology, 2025, 38(7): 58-65.

图/表 11

表1

特征值计算式"

特征值	计算式
小梯度优势	T₁=[ $∑ x ∑ y h (x, y) y 2$ ]/[ $∑ x ∑ y$ h(x,y)]
大梯度优势	T₂=[ $∑ x ∑ y$ h(x,y)j²]/[ $∑ x ∑ y$ h(x,y)]
灰度不均匀性	$T_{3}=\left\{\sum_{x}\left[\sum_{y} \boldsymbol{h}(x, y)\right]^{2}\right\} /\left[\sum_{x} \sum_{y} \boldsymbol{h}(x, y)\right]$
梯度不均匀性	$T_{4}=\left\{\sum_{y}\left[\sum_{x} \boldsymbol{h}(x, y)\right]^{2}\right\} /\left[\sum_{x} \sum_{y} \boldsymbol{h}(x, y)\right]$
能量	T₅= $∑ x ∑ y$ [h'(x,y)]²
平均灰度	T₆= $∑ x$ x[ $∑ y$ h'(x,y)]
平均梯度	T₇= $∑ y$ y[ $∑ x$ h'(x,y)]
灰度均方差	$T_{8}=\left\{\sum_{x}\left(x-T_{6}\right)^{2}\left[\sum_{y} \boldsymbol{h}^{\prime}(x, y)\right]\right\}^{1 / 2}$
梯度均方差	$T_{9}=\left\{\sum_{y}\left(y-T_{7}\right)^{2}\left[\sum_{x} \boldsymbol{h}^{\prime}(x, y)\right]\right\}^{1 / 2}$
相关度	T₁₀= $∑ x ∑ y$ (x-T₆)(y-T₇)h'(x,y)
灰度熵	T₁₁=- $∑ x ∑ y$ h'(x,y)log $∑ y$ h'(x,y)
梯度熵	T₁₂=- $∑ x ∑ y$ h'(x,y)log $∑ x$ h'(x,y)
混合熵	T₁₃=- $∑ x ∑ y$ h'(x,y)logh'(x,y)
惯性	T₁₄= $∑ x ∑ y$ (x-y)²h'(x,y)
逆差距	T₁₅= $∑ x ∑ y h' (x, y) 1 + (x - y) 2$

表1

图1

图2

图3

图4

图5

图6

图7

图8

表2

图9

参考文献 18

[1]	程锦锋, 方贵盛, 高惠芳. 表面缺陷检测的机器视觉技术研究进展[J]. 计算机应用研究, 2023, 40(4):967-977.
	Cheng Jinfeng, Fang Guisheng, Gao Huifang. Research progress of surface defect detection based on machine vision technology[J]. Application Research of Computers, 2023, 40(4):967-977.
[2]	马振华. 基于机器视觉的织物瑕疵检测方法[D]. 徐州: 中国矿业大学, 2023:10-20.
	Ma Zhenhua. Fabric defect detection method based on machine vision[D]. Xuzhou: China University of Mining and Technology, 2023:10-20.
[3]	丁光熊. 布匹瑕疵分类算法的研究[D]. 上海: 上海师范大学, 2020:1-30.
	Ding Guangxiong. Research on improvement on fabric anomaly classification[D]. Shanghai: Shanghai Normal University, 2020:1-30.
[4]	Zhuang E X, Wang W B. Escalator foundation bolt loosening fault recognition based on empirical wavelet transform and multi-scale gray-gradient co-occurrence matrix[J]. Sensors, 2023, 23(15):6801-6809.
[5]	王鹤翔. 基于小波变换和SVM的织物疵点检测与分类[D]. 上海: 东华大学, 2015:1-21.
	Wang Hexiang. Fabric defect detection and classification based on wavelet transformation and SVM[D]. Shanghai: Donghua University, 2015:1-21.
[6]	吴哲. 基于小波分析与 SVM 的织物疵点识别分类算法的研究[D]. 泉州: 华侨大学, 2016:10-34.
	Wu Zhe. Research on fabric defect recognition and classification algorithms based on wavelet analysis and support vector machine[D]. Quanzhou: Huaqiao University, 2016:10-34.
[7]	张秋元, 刘志全. 一种基于纹理特征分析的地铁隧道裂缝检测方案[J]. 电子科技, 2022, 35(11):36-41.
	Zhang Qiuyuan, Liu Zhiquan. A texture feature analysis-based crack detection scheme for metro tunnels[J]. Electronic Science and Technology, 2022, 35(11):36-41.
[8]	朱浩. 一种基于机器视觉的织物瑕疵检测系统研究[D]. 广州: 华南理工大学, 2021:1-76.
	Zhu Hao. A study of computer-vision-based fabric defect detection system[D]. Guangzhou: South China University of Technology, 2021:1-76.
[9]	陈滔, 张庆国, 刘澳. 基于灰度共生矩阵的图形纹理检测及焊接缺陷的SVM分类实现[J]. 洛阳理工学院学报(自然科学版), 2022, 32(1):53-61,67.
	Chen Tao, Zhang Qingguo, Liu Ao. SVM classification implementation of graphic texture detection and welding defects based on gray level co-occurrence matrix[J]. Journal of Luoyang Institute of Science and Technology(Natural Science Edition), 2022, 32(1):53-61,67.
[10]	田志新, 徐震, 茅健, 等. 基于多尺度特征融合的钢材表面缺陷分类方法[J]. 电子科技, 2024, 37(2):87-94.
	Tian Zhixin, Xu Zhen, Mao Jian, et al. Classification method of steel surface defects based on multi-scale feature fusion[J]. Electronic Science and Technology, 2024, 37(2):87-94.
[11]	Shah A F. Gabor frames on local fields of positive characteristic[J]. Tbilisi Mathematical Journal, 2016, 9(2):129-139.
[12]	Yildiz K, Buldu A, Demetgul M, et al. A novel thermal-based fabric defect detection technique[J]. The Journal of the Textile Institute, 2015, 106(3):275-283.
[13]	戴光, 崔巍, 张颖, 等. 基于灰度-梯度共生矩阵的焊缝缺陷聚类分析[J]. 中国安全科学学报, 2013, 23(3):79-85.
	Dai Guang, Cui Wei, Zhang Ying, et al. Cluster analysis of weld defects based on gray-gradient co-occurrence matrix[J]. China Safety Science Journal, 2013, 23(3):79-85.
[14]	邓秀勤, 熊勇, 彭宏. 一种有效的自适应加权中值滤波算法[J]. 计算机工程与应用, 2009, 45(35):186-187.
	Deng Xiuqin, Xiong Yong, Peng Hong. Effective adaptive weighted median filter algorithm[J]. Computer Engineering and Applications, 2009, 45(35):186-187.
[15]	Liu Y X, Wei H F, Zhang H J, et al. An efficient stochastic gradient descent algorithm to maximize the coverage of cellular networks[J]. IEEE Transactions on Wireless Communications, 2019, 18(7):3424-3436.
[16]	Fu Y, Wang Y W, Wang W Q. Content-based natural image classification and retrieval using SVM[J]. Chinese Journal of Computers, 2003, 26(10):1261-1265.
[17]	张黎, 王振全. 基于小波分析与支持向量机控制图混合模式识别[J]. 郑州航空工业管理学院学报, 2023, 41(4):64-72.
	Zhang Li, Wang Zhenquan. A study on pattern recognition based on wavelet analysis and support vector machine for hybrid of control charts[J]. Journal of Zhengzhou University of Aeronautics, 2023, 41(4):64-72.
[18]	肖靓. 基于支持向量机的图像分类研究[D]. 上海: 同济大学, 2006:1-48.
	Xiao Liang. Support vector machine-based image classification research[D]. Shanghai: Tongji University, 2006:1-48.

样本数量	GLCM/ms·f^-1	GGCM/ms·f^-1
训练样本数(2 553)	39.2	14.9
测试样本数(1 121)	38.4	14.3
总样本数(3 674)	39.0	14.7

基于灰度梯度共生矩阵的快速布匹瑕疵检测算法

A Fast Fabric Defect Detection Algorithm Based on Gray Gradient Co-Occurrence Matrix

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 18

相关文章 15

Metrics

本文评价

推荐阅读 0

[1]	莫松楠, 金海. 一种结合点、线、面特征的RGB-D SLAM算法[J]. 电子科技, 2025, 38(8): 66-72.
[2]	竹小锋, 白兴宇. 一种基于支持向量机的宽带声谱故障监测方法[J]. 电子科技, 2025, 38(6): 39-44.
[3]	胡佳炜, 王亚刚. 基于机器学习的助行机器人用户跌倒检测算法[J]. 电子科技, 2025, 38(4): 39-45.
[4]	田志新,徐震,茅健,林彬彬,廖薇. 基于多尺度特征融合的钢材表面缺陷分类方法[J]. 电子科技, 2024, 37(2): 87-95.
[5]	卢红文,袁旭峰,陈瑞洁,李雨龙. 柔性接入微电网的小信号建模[J]. 电子科技, 2023, 36(3): 21-28.
[6]	王桥,胡春燕,李菲菲. 基于深度迁移学习与多尺度特征融合的场景识别方法[J]. 电子科技, 2023, 36(11): 19-27.
[7]	邓源,施一萍,江悦莹,朱亚梅,刘瑾. 基于MobileNetV2与LBP特征融合的婴幼儿表情识别算法[J]. 电子科技, 2022, 35(8): 47-52.
[8]	黄元生,姜雨晴,王静. 基于改进TOPSIS-PSO-SVM的配电网适应性评价研究[J]. 电子科技, 2022, 35(6): 54-63.
[9]	伍伟嘉,杨俭,袁天辰,邵志慧. 基于支持向量机的轨道结构病害识别研究[J]. 电子科技, 2022, 35(2): 27-33.
[10]	邵志慧,杨俭,袁天辰,伍伟嘉. 基于排列熵和支持向量机的轨枕病害诊断[J]. 电子科技, 2022, 35(2): 52-58.
[11]	谢爱平,张雨生,刘莹,何梓昂,高锐. 基于瞬时特征参数和功率谱熵的联合调制识别[J]. 电子科技, 2022, 35(11): 104-110.
[12]	张秋元,刘志全. 一种基于纹理特征分析的地铁隧道裂缝检测方案[J]. 电子科技, 2022, 35(11): 36-41.
[13]	冯先丁,魏镜弢,吴张永,钱杰,浦友尚. 基于PCA-PSO-SVM的球磨机负荷预测研究[J]. 电子科技, 2022, 35(1): 29-34.
[14]	张朋飞,魏存伟,刘先康,刘安然,杨欧,林永霖. 窄带舰船目标识别方法研究[J]. 电子科技, 2021, 34(6): 50-55.
[15]	刘姝,邵杰,张颐婷,张善章. 基于双谱和谱特征的心电信号分类方法[J]. 电子科技, 2021, 34(5): 42-46.