基于学习的Gabor滤波器多样式布匹瑕疵检测

摘要

Complex cloth patterns make it difficult to detect defects in the manufacturing process .Thus ,a learning-based Gabor filter in multi-style fabric defect detection is proposed . First ,Gabor filters are used to filter multi-style cloth at different directions and scales and se-lect the optimal filter scale .Then ,the defect features are extracted by using digital image fea-ture discriminant ,the feature images are segmented ,and a number of preferred feature infor-mation blocks are merged .Finally ,the fused images are processed by binarized and morpholog-ical filtering .Experimental results show that the method performs effectively both in flaw de- tecting of multi-style cloth and in real-time performing .Compared with similar algorithms ,the detection accuracy and time efficiency have a certain increase .%由于布匹的花纹样式多,图案复杂,导致瑕疵检测困难等问题,给出了一种基于学习的Gabor滤波器实现多样式布匹瑕疵的检测方法.该方法首先利用多方向 、多尺度Gabor滤波器对多样式布匹图像滤波,优选最佳滤波尺度.然后通过判别式提取瑕疵特征,对特征图像分块处理,融合优选的特征信息块,最后对融合后的图像进行二值化和形态学滤波处理.实验结果表明,该方法能够有效检测多样式布匹瑕疵,具有较好的实时性.相比于同类算法,检测精度和时间效率均有一定的提高.