织物缺陷的自动检测是纺织行业所面临的技术难题之一。为了对织物缺陷进行快速准确的自动检测，建立了一种基于Gabor滤波器的织物缺陷自动检测方法，即首先根据无缺陷织物的结构特征，应用小生境遗传算法寻找最优的Gabor滤波器参数和分割阈值；然后将待检图像通过滤波和分割来得到检测结果，同时根据响应矩阵的极小值点来计算缺陷的形状特征。该方法应用于帘子布的缺陷检测的实验结果证明，该方法是可行的和有效的，而且还具有适用性广、识别能力强、检测速度快等特点。

The fabric defect automatic inspection is one of the technical problems that the textile industry is facing. This paper brings forward a approach for textile flaw automatic detection based on optimal Gabor filters. At first the niche genetic algorithm is employed to attain the parameters of the optimal Gabor filters and segmentation threshold based on.the feature of the unflawed textile, and then apply the optimal Gabor filter and threshold on the fabric image to achieve the resnlts. Meanwhile, the shape characteristics of defects are also achieved from the distribution of minimums of response matrix. The approach is used to detect the flaw of cord fabrics. The test results demonstrate this method is efficient and practicable, and also characterized by wide applied range, strong capability of recognition and high testing speed.