基于机器视觉的烟梗长度测量方法
Measurement method of tobacco stem length based on machine vision
董世超 1陈丙三 1连长伟 2卢敏瑞 2彭小冬 1张腾健 2王芳2
作者信息
- 1. 福建理工大学 机械与汽车工程学院,福建 福州 350118
- 2. 福建武夷烟叶有限公司,福建 邵武 354000
- 折叠
摘要
针对传统人工烟梗长度测量方法存在效率低、劳动强度大、精度难以保证等问题,提出了一种利用机器视觉测量不同形态烟梗长度的分类测量法.该方法通过2 500 万像素的高端面阵黑白相机采集烟梗清晰数字图像.首先,通过中值滤波结合图像开运算、全局阈值分割、图像连通性操作等步骤对图像进行预处理并提取烟梗感兴趣区域.然后,根据矩形度将其分为矩形和异形两种类型,以提高测量的准确性.最后,采用基于凸包获取最小外接矩形测量法和基于骨架提取自适应宽度的改进Steger算法,分别对两种形态的烟梗进行测量.选取15组长度不同的矩形梗和异形梗,通过三坐标测量仪测量烟梗的标准值与本方法测量值对比.结果表明:相较于骨架提取法和传统Steger算法,采用该方法对矩形与异形梗测量的最大相对误差分别为-0.196%和-0.109%,对单根烟梗的重复测量最大误差分别为 0.008 mm和0.006 mm,不但具有较高的精度和较好的稳定性,还可显著降低劳动强度,具备广泛的实际应用潜力.
Abstract
In response to the challenges posed by traditional manual measurements of tobacco stem length,char-acterized by low efficiency,high labor intensity,and difficulty in ensuring accuracy,a classification measure-ment method for measuring the length of different forms of tobacco stems by machine vision was proposed.This method collects clear digital images of tobacco stems through a 25 million pixel high-end array black-and-white camera.Initially,the images were subjected to preprocessing through median filtering and image opening opera-tions.Subsequently,global threshold segmentation and image connectivity techniques were applied to extract and crop the region of interest within the tobacco stem.Based on morphology,the tobacco stems were then categorized into rectangular and curved types to enhance measurement accuracy.To perform measurements,two distinct approa-ches were employed for these two morphological categories:the minimum bounding rectangle measurement method for rectangular stems and an improved Steger algorithm based on the adaptive width skeleton extraction for curved stems.Fifteen groups of rectangular stems and curved ones with different lengths were selected,and the standard values of tobacco stems measured by three-coordinate measuring instrument were compared with the measured values of this method.Experimental results demonstrate that the maximum relative errors of rectangular and irregular stems measured by this method are-0.196%and-0.109%,respectively,and the maximum repeated measurement errors for a single tobacco stem are 0.008 mm and 0.006 mm,respectively.This method not only has high accuracy and good stability,but also significantly reduces labor intensity and has broad practical application potential.
关键词
机器视觉/烟梗测量/最小外接矩形/改进Steger算法Key words
machine vision/tobacco stem measurement/minimum bounding rectangle/improved Steger algo-rithm引用本文复制引用
基金项目
中国烟草总公司福建省公司科技项目(2023350000200090)
出版年
2024
NETL
NSTL
万方数据