摘要
对流水线上工件外观尺寸的测量,往往难以保证测量结果的精度和算法的稳定性。借鉴有关研究成果,提出了一种基于随机森林分类器与图像分割技术的工件外观尺寸测量算法。首先采用传统图像处理方法,融合滤波、灰度化分割、形态学分析与连通区域特征分析,构建图像处理算子,得到工件的位置区域。然后结合多特征向量数据表,根据随机森林特性建立分类器,准确定位工件,进而提取测量结果。基于OpenCV开源函数库与C++图像函数来实现算法,并开发了测量软件。实验测试结果显示,按此算法,可以实现对待测工件的正确定位,且能够比较精确地测量出目标工件的尺寸。
As the accuracy and repeatability cannot be guaranteed in mass operation on assembly line,a workpiece appearance measurement system based on random forest and image processing is proposed in this paper.Firstly,the traditional image processing method is adopted,which combines filtering,gray segmentation,morphological analysis and feature analysis of connected area to form image processing operator and get the position region of the workpiece.Then,combined with multi-eigenvector data table,according to the random forest characteristics,a classifier is established to accurately locate the workpiece,and then extract the measurement results.Finally,based on OpenCV open source function library and C++image function,the measurement software is developed.The experimental results show that the proposed technology can not only locate the workpiece position,but also accurately measure the size of the target.
作者
占海文
ZHAN Haiwen(Anhui Sanlian University,Anhui 230601,China)
出处
《重庆科技学院学报（自然科学版）》
CAS
2020年第1期106-109,共4页
Journal of Chongqing University of Science and Technology：Natural Sciences Edition
基金
安徽省2017年高等学校省级质量工程项目“网络工程核心课程教学团队”(2017jxtd131)。
作者简介
占海文(1979—),男,硕士,讲师,政工师,研究方向为计算机网络及信息技术、高等教育管理。
