基于信息熵的喷水织机织物疵点检测  

Detection of fabric defects on water jet loom based on information entropy

作　　者：柏顺伟 沈丹峰[1] 赵刚 尚国飞 付茂文 BAI Shunwei;SHEN Danfeng;ZHAO Gang;SHANG Guofei;FU Maowen(School of Mechanical and Electrical Engineering,Xi’an Polytechnic University,Xi’an 710048,China;Shaanxi Changling TextileMechanical&Electronic Technological Co.,Ltd.,Baoji 721013,China)

机构地区：[1]西安工程大学机电工程学院,西安710048 [2]陕西长岭纺织机电科技有限公司,陕西宝鸡721013

出　　处：《丝绸》2022年第6期72-79,共8页Journal of Silk

基　　金：国家自然科学基金项目(51805402)。

摘　　要：为实现喷水织机织物上疵点的有效检测及疵点准确定位,本文提出一种基于二维信息熵的特征显著图的疵点检测方法。将待检测织物图像使用改进的同态滤波进行预处理,改善图像因光照不均对疵点检测产生的影响;利用图像基元信息熵与图像纹理的关系引入二维熵来反映图像纹理的空间分布,计算每个重叠的图像基元的信息熵,并把该熵值作为中心像素的灰度值,经归一化后生成一幅特征显著图;最后对显著图进行阈值分割得到疵点的轮廓,同时通过显著图得到径向投影差分序列对织物图像进行疵点判别。结果表明,本方法能够有效地抑制织物中重复的纹理背景,突出疵点部分,实现疵点区域的准确定位。In the production of textile industry,fabric defects caused by mechanical failure,yarn breakage,etc.occur from time to time,causing huge economic losses to enterprises,so fabric defect detection is a key link to ensure the quality of fabrics.Now,the textile industry has defined more than 70 kinds of fabric defects,and realizing the effective detection of fabric defects is the research hot spot in the field of machine vision because of the diversity of images and the complexity of production environment.Traditionally,most of the fabric defects are detected manually,but the success rate of manual detection is only 70%-80%,and the long working hours detrimental to workers’health and the production cost is high.In order to solve the problems brought by manual detection,it is important to realize the automatic detection of fabric defects by fast and reliable machine vision technology based on digital image processing.To realize the effective detection and accurate location of fabric defects on water jet loom fabric,a defect detection technique is proposed based on a two-dimensional information entropy feature salient map.Firstly,the fabric image to be detected is pre-processed with improved homomorphic filtering to improve the effect of unequal lighting on the fabric image caused by the production environment on the detection of defects,and to obtain a fabric image with consistent lighting.Then,the two-dimensional entropy is reflected in the spatial distribution of the image texture by using the correlation between the information entropy of the image primitive at the defect and the non-defective image primitive and the image texture,and the information entropy of each overlapping image primitive is calculated using a rectangular calculation kernel.The information entropy of each overlapping image primitive is calculated by using the rectangular calculation kernel,and the entropy value is used as the gray value of the central pixel,which is normalized to generate a significant map of defective features.Finally,a thr

关 键 词：喷水织机 同态滤波 信息熵 特征显著性 阈值分割 疵点检测 

分 类 号：TS106.5[轻工技术与工程—纺织工程] TP391.4[轻工技术与工程—纺织科学与工程]