摘要
为解决目前卷烟厂生产线对烟包缺支检测精度和速度要求高的问题,提出一种基于支持向量机(SVM)的烟包缺支在线检测方法。首先通过工业CCD摄像头采集烟包图像,并进行预处理;其次提取采集图像的灰度特征和纹理特征,根据相应指标构建特征空间;最后建立基于多项式和高斯核函数的SVM分类方法对烟包进行缺支检测。实验结果证明,所设计的方法有效,高斯核函数精度稍高,缺支检测准确率最高可达99.8%,检测算法平均消耗时间仅为120 ms,满足烟包缺支在线检测的精度和实时性要求。
In order to realize the high accuracy and high speed of cigarette package shortage detection,it proposes an online detection method based on support vector machine(SVM).It uses an industrial CCD camera to collect and deal with the image of cigarette packet,extracts the gray and texture features of the collected images,and constructs the feature space based on the corresponding index.Taking SVM classification method based on polynomial and Gaussian kernel functions,it detects the lack of cigarette packets.The experimental results show that the designed method is effective,the detection accuracy rate of the lack of cigarettes can reach 99.8%,and the detection time is 0.12 s,which meets the accuracy and real-time requirements of the online detection of the lack of cigarettes.
作者
邓春宁
Deng Chunning(Equipment Management Department,Longyan Tobacco Industry Co., Ltd., Fujian Longyan, 364000, China)
出处
《机械设计与制造工程》
2020年第3期77-80,共4页
Machine Design and Manufacturing Engineering
关键词
烟包缺支检测
预处理
灰度特征
纹理特征
支持向量机
cigarette packet detection
preprocessing
gray feature
texture feature
support vector machine