Method for detecting band carbide particles of bearing steel

By using the target detection model YOLOv8 and image processing algorithms OTSU, YEN, and ISODATA, combined with rating rules, the accuracy and efficiency of detecting the morphology of banded carbide particles in bearing steel have been improved, achieving highly accurate automated rating. This method is suitable for the detection and rating of banded carbides in bearing steel.

CN122115337APending Publication Date: 2026-05-29JIANGSU SUPERVISION & INSPECTION INST FOR PROD QUALITY
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
JIANGSU SUPERVISION & INSPECTION INST FOR PROD QUALITY
Filing Date
2026-01-30
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

The accuracy of detecting the morphology of banded carbide particles in bearing steel using existing technologies is low, which affects the accuracy of grading.

Method used

The YOLOv8 target detection model is used for initial detection. Combined with semantic segmentation, the widest band of carbides with boundaries and penetrating the image are selected for secondary magnification. The images are then binarized using the Otsu method, the YEN method, or the ISODATA iterative self-organizing data analysis algorithm. The degree of clustering and the contour area are calculated, and the images are rated according to the established rating rules.

Benefits of technology

It improves the accuracy of detecting the morphology of banded carbide particles in bearing steel, achieves an accuracy of over 90% for binarized carbide aggregation, enables automatic continuous processing and rating of batch images, shortens analysis time, and improves the objectivity and speed of rating.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122115337A_ABST
    Figure CN122115337A_ABST
Patent Text Reader

Abstract

The application provides a bearing steel belt carbide particle morphology detection method, which comprises the following steps: obtaining an original bearing steel belt carbide image, performing first amplification processing to obtain an amplified image; obtaining a target image by subjecting the amplified image to a target detection model; performing semantic segmentation on the target image to obtain a belt carbide image; selecting a belt carbide image with a boundary, penetrating the image, the widest width and the densest particles, performing secondary amplification processing to obtain a to-be-detected image; and performing particle morphology detection on the to-be-detected image to obtain a morphology detection result including an aggregation degree, a total contour area and an average contour area. The bearing steel belt carbide particle morphology detection method can improve the accuracy of bearing steel belt carbide particle morphology detection, can realize a binary carbide aggregation degree accuracy of greater than 90%, and can realize bearing steel belt carbide particle morphology detection with high accuracy.
Need to check novelty before this filing date? Find Prior Art