结合Otsu与EM的啤酒瓶图像分割及动态计数研究  

Beer Bottle Image Segmentation and Dynamic Counting Research Combined with Otsu and EM

作　　者：肖志云[1,2] 渠志云 XIAO Zhiyun;QU Zhiyun(College of Energy Power,Inner Mongolia University of Technology,Hohhot 010080,China;Inner Mongolia Key Laboratory of Electromechanical Control,Inner Mongolia University of Technology,Hohhot 010051,China)

机构地区：[1]内蒙古工业大学电力学院,呼和浩特010080 [2]内蒙古工业大学内蒙古机电控制重点实验室,呼和浩特010051

出　　处：《重庆理工大学学报（自然科学）》2020年第8期165-175,222,共12页Journal of Chongqing University of Technology：Natural Science

基　　金：国家自然科学基金项目(61661042)。

摘　　要：针对多道生产线啤酒瓶计数不准确的实际问题,提出了结合二维Otsu与EM法的啤酒瓶图像分割方法和实现了多道生产线动态计数算法。在研究中,对于图像分割部分,通过结合二维Otsu与EM算法对预处理后的序列啤酒瓶图像进行分割;对于生产线动态计数部分,在上述图像分割的基础上,利用hough变换和基于灰度特征的匹配追踪算法确定前后2张图像的端面中心和像素间距,比较前后2帧图像中已计和未计啤酒瓶,实现多道生产线啤酒瓶动态计数。通过对实际生产线采集的图像进行计数试验,结果表明:所提出的生产线动态计数方法准确率高达100%,能够较好地解决多道运输啤酒瓶计数不准确的问题。Aiming at the practical problem of inaccurate counting of beer bottles in multi-line production lines,combining two-dimensional Otsu method with EM algorithm,a beer bottle image segmentation method were proposed to realize dynamic counting algorithm of multi-channel production line.In the study,the part of the image segmentation,the preprocessed image was segmented by combining two-dimensional Otsu with EM algorithm.Then,based on the above image segmentation,for the dynamic counting part of the production line,the center of the end face and the image distance of the two images were determined by hough transformation and the matching method based on the gray feature.The dynamic counting of beer bottles in multi-channel production line was realized by comparing the counted and uncounted beer bottles of images before and after the two frames.Finally,through the experiments on the images collected by the actual beer production line,the results show that the accuracy of the proposed counting method is as high as 100%,which can better solve the problem of multi-channel beer bottles.

关 键 词：机器视觉 动态计数 啤酒瓶生产线 二维OTSU法 EM法 

分 类 号：TM391.41[电气工程—电机]