基于Hough变换与最小二乘拟合的香烟小包拉线错牙检测方法

Bracing Wires Dislocation Detection Method for Packet Cigarette Based on Hough Transform and Least Squares Fitting

针对小盒香烟包装生产过程中经常出现外包装透明薄膜上没有拉线或拉线错牙等现象,提出一种基于图像处理技术的检测方法。首先,对烟包拉线图像进行边缘检测,获得拉线的轮廓;其次,利用Hough变换检测小盒拉线的上下边缘;当Hough变换无法完全找到拉线的上下边界时,采用最小二乘算法对左右两侧拉线弧进行圆拟合,最后根据计算出的拉线宽度或两圆心的垂直距离判断拉线是否错牙并计算出拉线错牙偏移量。实验结果表明所提方法能有效地检测出烟包拉线的错牙程度,且具有较好的鲁棒性和有效性。

A detection method based on image processing was proposed for dealing with bracing wires dislocation or no bracing wires of the packet cigarette, which widely exits in cigarette production. First, edge detection is utilized to get the outline information for bracing wire pictures. Then, Hough Line transform is used to recognize the edge of bracing wires. If Hough transform cannot find all the edges of bracing wire, least squares circle fitting is applied to fit the bracing wire arcs. At last, whether the bracing wire is dislocate is determined and the offset is calculated according to the width of bracing wire or distance of the center of the circles. Experimental results show that the proposed method can detect the extent of the packaged cigarette bracing wires dislocation effectively and has a better robustness and effectiveness.