纹理织物疵点窗口跳步形态学法检测

针对纹理织物疵点自动检测时因生产速度快造成的织物抖动以及检测速度难以匹配问题,提出窗口跳步形态学法纹理织物疵点检测算法。使用该算法对图像进行窗口分割及预处理后,首先对纹理织物图像的纹理特征进行分析,然后设计形态学算子进行腐蚀操作,最后使用连通域分析来确定疵点大小及位置。仿真实验及工厂实际应用表明,该算法可有效克服工业生产中纹理织物抖动造成的图像明暗不均,可检测出纹理织物中存在的破洞、经纬疵点、污渍、断线、折痕和结头等各种疵点,而且检测速度明显优于快速傅里叶变换特征点算法以及传统形态学检测算法。实时检测速度超过80 m/min,疵点检测精度为0.1 mm,满足实际生产需求。

Textured fabric defect detection based on windowed hop-step morphological algorithm

Aim at the problem of low detection efficiency and fabric jittering due to high production rate, when texture fabric defects are automatically detected. A textured fabric defect detection was presented based on a windowed hop-step morphological algorithm. Firstly, window segmentation and preprocessing on images were carried out, and then the image texture features of the textured fabric were analyzed. Secondly morphological operators were designed for corrosion operation. Finally, the defect size and location were determined by connected domain analysis. Experimental simulation and practical application results show that the algorithm can solve the problem of the images of uneven light and shade caused by the cloth trembling effectively, and the algorithm can detect the presence of defects in the fabric texture including broke holes, warp and weft defects, stains broken lines, creases, knots and so on. The detection algorithm has high stability and reliability, thus can meet the actual production demand. The detection speed is superior to the ( fast fourier transform algorithm) feature point algorithm and conventional morphological algorithm. The real-time detection speed is over 80 m/min, and the size of the defect detection accuracy is 0. 1 mm.