基于直方图均衡化的毛织物服装印花缺陷检测  

Detection of textile printing defects based on histogram equalization

作　　者：张玉芹[1] 杨文明[2] ZHANG Yuqin;YANG Wenming(Sino-UK College of Fashion and Design,Zhejiang Fashion Institute of Technology,Ningbo,Zhejiang 315211,China;School of Design&Communication,Zhejiang Fashion Institute of Technology,Ningbo,Zhejiang 315211,China)

机构地区：[1]浙江纺织服装职业技术学院中英时尚设计学院,浙江宁波315211 [2]浙江纺织服装职业技术学院设计媒体学院,浙江宁波315211

出　　处：《毛纺科技》2024年第6期89-95,共7页Wool Textile Journal

基　　金：“纺织之光”中国纺织工业联合会职业教育教学改革项目(2020ZJJGLX098)。

摘　　要：为了保证毛织物印花服装的生产质量,提出基于直方图均衡化的毛织物服装印花缺陷检测方法。设置毛织物服装印花不同缺陷特征作为检测标准,利用光学成像原理采集毛织物服装印花图像,通过颜色转换、图像滤波等步骤,实现初始印花图像的预处理;利用直方图均衡化技术提取服装印花图像特征,通过特征匹配确定缺陷状态与类型,实现毛织物服装印花缺陷的检测。测试结果表明,优化设计方法得出缺陷面积检测误差的平均值为0.09 mm^(2),缺陷类型检测错误率较低。In order to ensure the production quality of woollen fabric printed clothing,a method of detection of woollen fabric printed defect based on histogram equalization was proposed.The different defect features of woollen garment printing were set as the detection criteria,the woollen garment printing images were collected by optical imaging principle,and the initial printing images were preprocessed by color conversion and image filtering.Histogram equalization technology is used to extract the features of clothing printing images,and the defect state and type are determined by feature matching to realize the detection of wool clothing printing defects.The test results show that the average error of defect area detection is 0.09 mm^(2),and the error rate of defect type detection is low.

关 键 词：直方图均衡化 毛织物服装 服装印花缺陷 缺陷检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]