SLEE算法与Hough圆拟合在香烟小包拉线错牙检测中的研究

Research on SlEE Algorithm and Hough Circle Fitting for Cigarete Bracing Wires Dislocation Detection

摘要:

针对香烟生产中广泛存在的小包拉线错牙问题,提出一种基于图形识别的检测方法；利用图像校正、平滑滤波、迭代阈值分割、边缘提取对拉线图像进行预处理,再采用随机霍夫变换(RHT)对两个拉线U型切口进行圆拟合,进而根据两个圆心在垂直方向上的距离计算出拉线错牙偏移量；针对Sobel、Canny等获取的边缘存在较多冗余信息问题,提出了一种扫描线边缘提取(SLEE)算法；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,误差小于0.3 mm且具有较好的鲁棒性。

Abstract:

A detection method based on shape recognition is proposed for dealing with the cigarette bracing wires dislocation problem, which widely exits in cigarette production. Firstly, image correction, smoothing filtering, iterative threshold segmentation and edge extraction are utilized to preprocess the image of cigarette bracing wires. Secondly, the random Hough transform is used to recognize two U-type incisions of bracing wires and fit a circle to each one, then the distance of two circle centers in the vertical direction is obtained as the offset of bracing wires. A scanning line edge extraction(SLEE) algorithm is put forward for the issue that the edge has more redundant information acquired by Sobel and Canny. Experimental results show that the proposed method can detect the extent of the packaged cigarette bracing wires dislocation effectively, the errors are less than 0.3mm and has a better robustness.