基于ButterWorth滤波和EM算法的电子布疵点检测

为了精确定位和分割电子布生产过程中产生的疵点，提出一种基于ButterWorth滤波和EM算法的电子布疵点检测方法。因电子布的背景纹理对疵点检测存在影响，首先采用ButterWorth滤波弱化图像背景纹理信息；再应用高斯混合模型对预处理后的图像进行表征,通过EM算法迭代模型的最优解，对像素进行标记；最后，根据标记结果分离背景与疵点区域，定位并分割疵点。实验结果表明，该方法既能准确定位疵点位置，又能有效保留疵点的细节信息，并且能够检测多种类型的疵点，在疵点检测方面具有一定的参考价值。

Glass Fiber Fabric Defect Detection Based on ButterWorth Filter and EM Algorithm

In order to accurately locate and segment the defects generated in the production process of glass fiber fabric,an defect detection method based on Butterworth filtering and EM algorithm is proposed.Because the background texture of the glass fiber fabric has great impact on the defect detection,firstly,ButterWorth filtering is used to weaken the background texture information of the image,then the Gaussian mixture model is used to characterize the pre-processed image,and EM algorithm is used to iterate the optimal solution of the model to mark the pixels.Finally,the background and the defect area are separated according to the marked results,and the defects are located and segmented.The experimental results show that the proposed algorithm can not only accurately locate the defect,but also effectively retain the details of the defects,and can detect many types of defects,which has certain reference value in defect detection.