A Study on High-Precision Dimensional Measurement of Irregularly Shaped Carbonitrided 820CrMnTi Components

Area of Science:

• Materials Science and Engineering
• Computer Vision and Image Processing
• Metrology

Background:

• Industrial inspection of irregularly shaped 820CrMnTi carburizing and nitriding parts faces challenges including overexposure due to high reflectivity, low surface contrast, and interference from burrs.
• Carburizing and nitriding heat treatments alter surface photovoltaic characteristics, complicating visual inspection under variable industrial lighting conditions.
• Traditional inspection methods are susceptible to local overexposure and false out-of-tolerance measurements caused by outlier sensitivity.

Purpose of the Study:

• To develop an innovative precision detection method for industrial online inspection of irregularly shaped, highly reflective parts.
• To address challenges such as overexposure, low contrast, and burr interference in automated visual inspection systems.
• To improve the accuracy and robustness of inspection for parts subjected to carburizing and nitriding treatments.

Main Methods:

• An adaptive imaging technique integrating a dual-drive heterogeneous coupling model (RGFCN) was developed.
• A programmatic adaptive exposure control algorithm based on grayscale histogram feedback was implemented for real-time parameter adjustment.
• An adaptive main-axis scanning strategy, Gaussian gradient energy fields, Huber M-estimation, and gradient boosting decision trees (GBDTs) were employed for noise suppression, defect reduction, and error compensation.

Main Results:

• The proposed method effectively suppresses overexposure and reduces projection errors through adaptive scanning.
• Robust fitting mechanisms significantly reduced interference from burrs and oil stains, achieving immunity to non-functional defects.
• The system demonstrated excellent robustness in simulated factory conditions with lighting fluctuations and oil stains, reducing the false out-of-tolerance rate by over 90% and achieving micrometer-level measurement repeatability.

Conclusions:

• The integrated adaptive imaging and RGFCN model provides an effective solution for the visual inspection of complex, irregular, and highly reflective industrial parts.
• The method significantly enhances inspection reliability by mitigating issues related to surface properties and environmental interference.
• This approach offers a robust and accurate solution for quality control on dynamic production lines, meeting stringent metrological requirements.