皮革科学与工程 ›› 2023, Vol. 33 ›› Issue (1): 26-31.doi: 10.19677/j.issn.1004-7964.2023.01.005

• 试验研究 • 上一篇    下一篇

基于计算机视觉技术的皮革边缘抓取点的定位

陈传旗1, 张辉1, 强西怀1,2,*, 樊小蒲3   

  1. 1.陕西科技大学轻工科学与工程学院,陕西 西安 710021;
    2.轻化工程国家级实验教学示范中心,陕西 西安 710021;
    3.陕西科技大学机电工程学院,陕西 西安 710021
  • 收稿日期:2022-07-15 出版日期:2023-02-01 上线日期:2023-01-04
  • 通讯作者: *强西怀(1963-),男,教授,研究方向为清洁化制革生产。E-mail:qiangxihuai@163.com。
  • 作者简介:陈传旗(1992-),男,硕士研究生,研究方向为轻工机械自动化。E-mail: a191105746@gmail.com。
  • 基金资助:
    陕西科技大学自然科学研究基金(2017BJ-59)

Positioning of Leather Edge Grab Points based on Computer Vision Technology

CHEN Chuanqi1, ZHANG Hui1, QIANG Xihuai1,2,*, FAN Xiaopu3   

  1. 1. College of Bioresources Chemical and Materials Engineering, Shaanxi University of Science and Technology, Xi’an 710021, China;
    2. National Demonstration Center for Experimental Light Chemistry Engineering Education, Xi’an 710021, China;
    3. Electrical and Mechanical Engineering College of Shaanxi University of Science and Technology,Xi’an 710021,China
  • Received:2022-07-15 Online:2023-02-01 Published:2023-01-04

摘要: 皮革边缘抓取点的准确定位是实现皮革自动绷板的关键环节。在皮革自动绷板过程中,为了实现对形状不规则和摆放位置不固定的革坯准确抓取,提出了基于计算机视觉技术对于革坯边缘抓取点进行定位的技术思路。首先利用棋盘格对相机进行标定,通过对采集的图像进行预处理,提取图像单像素轮廓,然后运用坐标系转换和轮廓点—质心角度算法筛选出革坯目标抓取点,最后结合标定后的相机像素精度,计算抓取点和质心的实际距离。方法平均运行耗时32.48 ms,平均定位角度误差0.16°,平均定位距离误差0.56 mm,这表明本方法可以准确实现对革坯边缘抓取点的定位,满足革坯自动绷板定位的需求。

关键词: 计算机视觉, 皮革自动绷板, 视觉定位, 坐标系转换

Abstract: The accurate positioning of the leather edge gripping point is a key aspect of automatic leather toggling. In order to achieve accurate gripping of irregularly shaped and irregularly placed crust leather, this paper proposes a method for locating leather edge gripping points based on computer vision technology in the automatic crust leather toggling process. The camera is first calibrated using a tessellation grid, and the single-pixel contour of the image is extracted by pre-processing the acquired image, then the target gripping point is selected using the coordinate system conversion and contour point - centre of mass angle algorithm, and finally the actual distance between the gripping point and the centre of mass is calculated by combining the calibrated camera pixel accuracy. The experimental results show that the average running time of this method is 32.48 ms, the average positioning angle error is 0.16°, and the average positioning distance error is 0.56 mm, which indicates that this method can accurately achieve the positioning of leather edge grasping points and meet the demand of crust leather automatic toggle positioning.

Key words: computer vision, automatic leather toggling, image positioning, coordinate system conversion

中图分类号: