基于图像增强局部上采样SSD的直线电机动子非接触位置检测方法研究
DOI:
作者:
作者单位:

合肥工业大学电气工程与自动化学院

作者简介:

通讯作者:

中图分类号:

TP391.41

基金项目:

国家自然科学基金项目(面上项目,重点项目,重大项目)


Research on non-contact position detection method of linear motor mover based on image enhanced local upsampling SSD
Author:
Affiliation:

School of Electrical Engineering and Automation Engineering, Hefei University of Technology

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    本文研究了一种基于图像增强局部上采样平方差和(IE-LUSSD)的高精度亚像素检测算法, 以提高直线电机动子位置检测对不同光照强度的抗干扰能力. 首先, 根据直线电机一维刚体平移的运动特点, 设计了一种基于线阵相机和非周期栅栏图像的动子位置检测系统, 线阵相机固定在动子上并跟随动子移动采集信号序列. 其次,通过灰度线性变换图像增强算法对采集到的信号序列进行预处理以增强图像信息. 然后, 通过 SSD 算法获取相邻信号序列间的整像素位移, 为了进一步提高测量准确性, 采用频率域矩阵乘法离散傅里叶变换对相邻信号间相关函数的峰值邻域进行上采样细化峰值曲线. 最后通过搭建了动子位置检测的实验平台验证了本文方法对不同光照条件的适应性. 本文算法可以达到 0.01 pixels 的检测精度, 动子的实际位置检测误差在 0.025 mm 以内.

    Abstract:

    A high precision subpixel detection algorithm based on image enhanced local up-sampling square variance and (IE-LUSSD) is studied to improve the anti-interference ability of linear motor position detection to different light intensity. Firstly, according to the motion characteristics of one-dimensional rigid body translation of linear motor, a motion position detection system based on line-scan camera and aperiodic fence image is designed. The line-scan camera is fixed on the mover and follows mover motion to collect signal sequences. Secondly, the gray-scale linear transform image enhancement algorithm is used to preprocess the collected signal sequences to enhance the image information. Then, SSD algorithm is adopted to obtain the integer-pixel displacement between adjacent signal sequences. In order to further improve the measurement accuracy, frequency domain matrix multiplication discrete Fourier transform is applied to up-sample the peak neighborhood of the correlation function of adjacent signals to refine the peak curve. Finally, method effectiveness in different light conditions is verified by building an experimental platform for motion position detection. Detection accuracy of the proposed algorithm can reach 0.01 pixels and actual position detection error of the mover is within 0.025 mm.

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2021-10-16
  • 最后修改日期:2022-03-03
  • 录用日期:2022-03-15
  • 在线发布日期:
  • 出版日期: