轧光工序中织物接缝线的图像检测

针对轧光工艺中织物接缝线的自动检测，本文提出了一种基于数字图像处理的判别方法。以防羽布的轧光工艺为例，首先对获取的图像进行灰度化与中值滤波预处理；然后对预处理的图像提取均值、标准差与变异系数(CV值)等特征参数；最后通过对这三项特征参数的分析与对比，发现特征参数CV值可以作为防羽布有无接缝线的判别依据。实验结果表明，特征参数CV值能够精确地判别织物有无接缝线，并且图像处理的计算量小，检测速度快，可实现接缝线的及时避让。

Fabric seam detection based on image processing in calendaring process

To realize the fabric seam in calendaring process, a detection method is proposed to recognize the seam of image based on digital image processing. Taking down-proof fabric in calendaring process as example, the down-proof fabric images are processed into gray image and filtered with a median kernel firstly; and then the mean value, standard value and coefficient of variation (CV value) are extracted as the feature parameters; by comparing and analyzing the three feature parameters, CV value can be used to recognize the seam of fabric image at last. The experimental results show that the CV value is better suited to recognize the fabric seam which has a small amount of calculation with a high detection rate.