基于Gabor滤波器对织造前帘子布疵点检测研究  

Defect detection of cord fabric before weaving based on Gabor filter

作　　者：唐旭晟[1] 杨思材 陈丹[1] 

机构地区：[1]福州大学机械工程及自动化学院,福建福州350116

出　　处：《机电工程》2017年第7期802-805,共4页Journal of Mechanical & Electrical Engineering

基　　金：福建省自然科学基金资助项目(2013J01226)

摘　　要：针对织造前帘子布纱线疵点自动检测问题,对织造前帘子布纱线疵点分布特点开展了研究分析,提出了一种基于Gabor滤波器的帘子布纱线疵点检测算法,利用Gabor滤波器的能量值来描述帘子布纱线疵点特征。通过对疵点Gabor图像阈值化处理,得到了疵点的二值图像。然后对疵点二值图像沿水平方向进行了均等分割,得到了疵点所在的条状图像并除去随机噪声。在1 792个疵点图像测试集上对不同尺度参数的Gabor滤波器进行疵点检测准确度分析,建立了最优的方向和尺度参数组合。并搭建了基于机器视觉技术的帘子布纱线疵点检测系统。研究结果表明,该检测算法对帘子布纱线疵点检测准确度达到99.2%。经过4个月生产实践表明,该系统对织造前帘子布纱线疵点检测效果比人工检测效果提高86%,实现了机器换人的目的。Aiming at the problem of automatic detection of the pre-weaving cord fabric yarn defect,the analysis of the distribution features of the pre-weaving cord fabric yarn defect was conducted and the algorithm for the detection of the cord fabric yarn defect based on the Gabor filters was proposed. The energy value of the Gabor filter was used to describe the defect characteristics of the cord fabric yard. The defect Gabor images were given thresholding process and the binary images of the defect were obtained. Then the defect binary images were equally divided in the horizontal direction to obtain the defect stripe images for removing the random noise. After the accuracy analysis of the defect detection with the Gabor filters with different parameters performed on 1792 test sets of the defect images,the optimal portfolio of the direction and scale of the defect Gabor features was established and the automatic detection system for the cord fabric yarn defect based on the machine vision was constructed. The results indicate that using this method raises the cord fabric yarn＇s defect detection accuracy up to 99. 2%. After the actual practices over 4 months,it gives an improvement by 86% compared with the manual detection and achieves the purposes of machine substitutions.

关 键 词：帘子布纱线 织造前 GABOR 疵点检测 机器视觉 

分 类 号：TP2[自动化与计算机技术—检测技术与自动化装置] TS1[自动化与计算机技术—控制科学与工程]