曲线匹配算法及其在织物自动对花检测中的应用

Curve matching algorithm and its application in fabric automatic register detection

针对圆网印花过程中的对花检测问题,在基于机器视觉的圆网印花检测系统的基础上,提出利用曲线匹配策略得到错花套色的位置误差矢量的方法。为确保目标边缘图像与标准边缘图像位于同一坐标系下,首先利用Fourier-Mellin曲线匹配,对其进行匹配消除拍摄误差;然后用基于傅里叶变换位移定理的曲线匹配算法对调整好的目标边缘图像和标准边缘图像进行匹配,检测对花误差。该匹配策略消除了拍摄误差,数据运算量小。实验表明其检测精度和运算速度可以满足织物对花在线检测的要求。

In view of the register detection problem in the rotary screen printing process, the curve matching strategy was proposed to get the position vector of off-pattern-topping on the basis of the rotary screen printing machine automatic detection system via machine vision. In order to ensure that the target edge image and the standard edge image are located in the same coordinate system, Fourier-Mellin curve matching method was used to match them to eliminate the shooting error; and then the curve matching method based on the Fourier transform shift theorem was used to match the well-tuned target edge image and the standard edge image to detect the register error. This matching strategy eliminates the error, and the computation volume is small. The experiments verify that its detection accuracy and computing speed can meet the requirements of the fabric register detection on line.