应用遗传算法优化Gabor滤波器的机织物疵点检测

Woven Fabric Defect Detection Using Gabor Filter Optimized by Genetic Algorithm

针对传统人工检测机织物疵点时准确率低的问题,提出一种采用遗传算法优化Gabor滤波器的方法用于机织物疵点检测。该方法分为两个阶段:在学习阶段,采用遗传算法对构造的Gabor滤波器参数(f 0,B,λ)进行优化,得到一组优化参数;在检测阶段,应用该组优化参数构造的Gabor滤波器对待检测的织物图像进行滤波处理获得特征图像。将特征图像进行融合,并对融合后图像进行阈值分割,检测出该图像是否存在疵点。为验证该方法的有效性,对采集的60幅含有断经、粗纬和跳纱等几类常见疵点图像进行检测,并与其他算法检测效果进行对比。结果表明,该方法疵点检测准确率达到96.67%,且能更准确快速地检测出织物疵点,证实了该方法的有效性。

Aiming at the low accuracy of traditional manual detection of fabric defects,a genetic algorithm optimization Gabor filter is proposed for woven fabric defect detection.The method was divided into two stages:in the learning stage,the genetic algorithm was used to optimize the constructed Gabor filter parameters(fo,β,λ)to obtain a set of optimization parameters;in the detection stage,the Gabor filter constructed by the optimized parameters was applied to the fabric image to be detected.A filtering process was performed to obtain feature images,then the feature images was fused,and the fused image was subjected to threshold segmentation to detect whether the image had defects.In order to verify the effectiveness of the method,sixty kinds of common defect images including warp,coarse weft and jump yarn were detected and compared with other algorithms.The results show that the detection accuracy of the method reaches to 96.67%,and the fabric defects can be detected accurately and quickly.Thus the effectiveness of the method is clarified.