Scholarlink

<p>Silicon nitride bearing roller microcracks at the stress concentration produce local spalling, resulting in reduced service life. To efficiently and completely extract the microcracks of silicon nitride bearing rollers and reduce the leakage rate and false detection rate of machine vision intelligent detection. Aiming at the problem of incomplete identification of microcracks in silicon nitride bearing rollers due to the low contrast between microcracks and the image background, a machine vision intelligent inspection system for silicon nitride bearing rollers is constructed, a multi-scale spatial weight function equation is constructed to enhance the image contrast, a multi-scale bilateral filtering and adaptive OTSU threshold fusion method is proposed for the identification of microcracks, and a maximized interclass variance function equation is designed to realize the extraction of microcrack. The experimental results show that the average peak signal-to-noise ratio of the microcracks enhanced image of the silicon nitride bearing roller reaches 44.50&#xa0;dB, and the average volume correlation error of the microcrack extraction reaches 90.77%, which improves the accuracy of the microcrack extraction of the silicon nitride bearing roller.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Microcrack Extraction from Silicon Nitride Bearing Rollers Based on Multi-scale Bilateral Filtering with Adaptive Otsu Threshold Segmentation

  • 期刊论文

  • 发表日期：

    2025年9月8日

查看全文
  • Qi Zheng,
  • Nanxing Wu,
  • Hui Yang,
  • Yan Zhao,
  • Tao Chen,
  • Zengguang Lai

摘要

Silicon nitride bearing roller microcracks at the stress concentration produce local spalling, resulting in reduced service life. To efficiently and completely extract the microcracks of silicon nitride bearing rollers and reduce the leakage rate and false detection rate of machine vision intelligent detection. Aiming at the problem of incomplete identification of microcracks in silicon nitride bearing rollers due to the low contrast between microcracks and the image background, a machine vision intelligent inspection system for silicon nitride bearing rollers is constructed, a multi-scale spatial weight function equation is constructed to enhance the image contrast, a multi-scale bilateral filtering and adaptive OTSU threshold fusion method is proposed for the identification of microcracks, and a maximized interclass variance function equation is designed to realize the extraction of microcrack. The experimental results show that the average peak signal-to-noise ratio of the microcracks enhanced image of the silicon nitride bearing roller reaches 44.50 dB, and the average volume correlation error of the microcrack extraction reaches 90.77%, which improves the accuracy of the microcrack extraction of the silicon nitride bearing roller.