提出一种结合小波变换与共现矩阵用于纺织品图像缺陷检测的方法。首先将灰度图像分解成子带 然

后将纹理图像分割成互不重叠的子窗口, 提取共现特征 最后用无缺陷样品训练的M ahalanob is分类器将每一子

窗口划分为缺陷的和无缺陷的。应用该算法进行实际工厂环境中的纺织品缺陷检测。实验结果表明, 集中处理

具有强判决能力的某一频带提高了检测性能, 也改善了计算效率。-Propose a wavelet transform and co-occurrence matrix for the textile image defect detection method. First, the gray image is decomposed into sub-band　and then the texture image into non-overlapping sub-windows were now feature extraction　Finally, defect-free samples of M ahalanob is trained classifier to each child window is divided into defective and non- defects. Practical application of the algorithm textile factory defect detection in the environment. Experimental results show that the decision to focus with a strong ability to improve the detection performance of a band, but also improve the computation efficiency.