摘要
为提高复杂背景下织物疵点检测的准确率,提出了一种融合主结构提取和多尺度线性滤波的疵点检测方法。该方法对输入图像进行中值滤波和对数增强预处理以增加疵点和背景的对比度,并利用相对总变差模型提取织物图像的主结构信息——疵点,达到抑制背景纹理的效果;然后利用多尺度线性滤波器实现精确定位达到增强疵点区域的目的;最后利用预处理与疵点增强结果进行点乘的形态学处理得到完整疵点区域。实验结果表明:此方法可以检测污渍、破洞、带纱、霉斑、结团等多种疵点类型,与RPCA等3种检测方法相比,检测准确率提高了9%以上。
In order to improve the accuracy of fabric defect detection in complex background,a defect detection method based on main structure extraction and multi-scale linear filtering was proposed.The input image was preprocessed by median filtering and logarithmic enhancement to increase the contrast of the fault and the background,and the main structure information defect of fabric image was extracted by relative total variation model to suppress the background texture;Then the multi-scale linear filter was used to accurately locate and enhance the fault region;Finally a complete defect region was obtained by morphological processing of point multiplication based on preprocessing and defect enhancement results.The experimental results show that the method can detect stains,holes,mildew,knots,compared with RPCA and other three detection methods,the detection accuracy of the method is improved by more than 9%.
作者
陈雪阳
潘杨
朱磊
翟子豪
CHEN Xueyang;PAN Yang;ZHU Lei;ZHAI Zihao(School of Electronics and Information,Xi’an Polytechnic University,Xi’an 710048,China)
出处
《纺织高校基础科学学报》
CAS
2021年第4期53-61,共9页
Basic Sciences Journal of Textile Universities
基金
国家自然科学基金(61971339)
陕西省重点研发计划项目(2019GY-113)。
关键词
织物疵点
主结构提取
多尺度线性滤波
相对总变差模型
fabric defect
main structure extraction
multi-scale linear filtering
relative variation model
