Abstract: Fabric defect detection based on traditional optimal Gabor filter has the drawbacks of low accuracy and poor robustness. An improved selection method is proposed to choose the optimal Gabor filter. The method is based on a cost function which is the sum of squared mean difference of each blocks indefect image and standard image. A group of Gabor filters are selected and the relevant parameters are figured out from standard images. The improved selection method is used to choose the optimal filtered image. The optimal filtered image is segmented with iterative threshold segmentation. According to the pixel information of the segmented image, it is detected whether the defect is included and the defect information is output. The method is validated by experiments. Compared with the traditional optimization selection method, the results show that the method has less computational complexity and high inspection performance, which can meet the requirements of online detection.

Key words: fabric defect detection, optimal Gabor filter, cost function

摘要： 针对传统Gabor优化选择方法用于布匹瑕疵检测时准确率低、鲁棒性差的缺点，提出了改进的优化选择方法，通过瑕疵图像与标准图像Gabor滤波后分块子图均值差平方和的代价函数实现优化选择。设置一组不同方向和尺度的Gabor滤波器并提取标准图像滤波后相关参数，通过改进的优化选择方法实现滤波后瑕疵图像的最优选择，利用迭代式阈值分割对最优滤波后图像进行二值分割，根据分割后图像的像素信息检测是否含有瑕疵并输出瑕疵信息。实验验证该方法，并与传统优化选择方法对比分析，结果表明该方法运算量较少，且检查性能高，可满足在线检测要求。

关键词: 布匹瑕疵检测, Gabor优化选择, 代价函数