气泡高干扰在线铁谱视频图像的磨粒快速分割算法研究  

Study on Fast Wears Segmentation Algorithm for Online Ferrography Video Images in High Interference of Bubble

作　　者：李梦琳 范斌 刘勇 张鹏 杜文亮[1] 毛军红[3] LI Menglin;FAN Bin;LIU Yong;ZHANG Peng;DU Wenliang;MAO Junhong(School of Mechanical and Electrical Engineering,Inner Mongolia Agricultural University,Hohhot Inner Mongolia 010018,China;State Key Laboratory of Intelligent Manufacturing of Special Vehicles and Transmission Systems,Baotou Inner Mongolia 014030,China;School of Mechanical Engineering,Xi􀆳an Jiaotong University,Xi􀆳an Shaanxi 710049,China)

机构地区：[1]内蒙古农业大学机电工程学院,内蒙古呼和浩特010018 [2]特种车辆及其传动系统智能制造国家重点实验室,内蒙古包头014030 [3]西安交通大学机械工程学院,陕西西安710049

出　　处：《润滑与密封》2021年第11期96-102,共7页Lubrication Engineering

基　　金：国家自然科学基金地区科学基金项目(51965054);内蒙古自治区自然科学基金面向项目(2021MS05041);特种车辆及其传动系统智能制造国家重点实验室开放课题(GZ2019KF009);内蒙古农业大学高层次人才科研启动项目(NDYB2019⁃9)。

摘　　要：针对在线铁谱视频图像气泡高干扰所面临的磨粒分割困难问题,提出一种气泡高干扰在线铁谱视频图像的磨粒快速分割算法。首先运用运动检测的方法确定视频中气泡的位置,并用相邻帧相同位置的图像信息对气泡区域进行处理,再使用双边滤波对处理后的图像进行平滑去噪,实现气泡干扰的初步抑制;最后基于抑制气泡图像的灰度直方图,对每一帧图像选取其自适应的阈值,实现在线铁谱视频图像中磨粒的快速分割。该研究为在线铁谱的磨粒分割与后续对磨粒特征的智能提取和分析奠定了基础。A fast wear particle segmentation algorithm based on high bubble interference was proposed to solve the diffi⁃cult problem of wear particle segmentation in online ferrography video images.Firstly,the location of bubbles in the video was determined by motion detection method,and the area of bubbles was processed by the image information in the same position of adjacent frames.Then,the processed image was smoothed and denoised by bilateral filtering to realize the initial suppression of bubble interference.Finally,based on the gray histogram of the images after bubble elimination,the adaptive threshold of each frame was selected to realize the fast segmentation of wear particles in on⁃line ferrography video images.This study lays a foundation for on⁃line ferrography wear particle segmentation and the next intelligent extraction or analysis of wear particles characteristics.

关 键 词：磨粒分割 运动物体检测 气泡抑制 气泡高干扰 在线铁谱视频 

分 类 号：TH117.1[机械工程—机械设计及理论]