Zhuojin Pipe Fitting Co., LtdMachine Vision Based Flange Dimension Detection System for Automotive Pipe Fittings
Literature Overview
This 2023 paper from Hubei University of Automotive Technology presents the development of a machine vision-based dimension detection system for automotive pipe fitting flanges. The system uses HALCON software for camera calibration and image processing, and C# programming for result display and integration. The work was supported by the Ministry of Education Industry-Education Collaboration Program and the Shiyan City Science and Technology Research Program.
System Architecture and Methodology
The detection system addresses the limitations of traditional manual sampling inspection, which suffers from low efficiency and high false detection rates. The system architecture consists of three main components:
| Component | Function | Key Technology |
|---|---|---|
| Imaging system | Capture flange images | Industrial camera, lighting, lens |
| Processing system | Extract and measure dimensions | HALCON software, Canny edge detection |
| Output system | Display and report results | C# programming, data storage |
The image processing pipeline follows a structured approach:
- Camera calibration: The camera is calibrated using HALCON to establish the relationship between pixel coordinates and real-world dimensions.
- Image preprocessing: Median filtering is applied to remove noise, followed by grayscale conversion for image enhancement.
- Edge detection: The Canny operator is used for sub-pixel edge detection, providing high-precision edge localization.
- Contour fitting: The least squares method is applied to the sub-pixel edge points to fit a continuous aperture contour.
- Dimension extraction: The flange face dimensions are calculated from the fitted contour.
- Result display: The measurement results are displayed and recorded through the C# interface.
Performance and Validation
The system achieves a measurement accuracy of 0.08 mm, which meets the dimensional requirements for automotive pipe fitting flanges. The detection time per flange is 1.2 seconds, representing a 70% improvement in efficiency compared to traditional manual inspection. The system was validated by comparing its measurements with those from a coordinate measuring machine, and the results showed good agreement within the stated accuracy.
| Performance Metric | Machine Vision System | Manual Inspection | Improvement |
|---|---|---|---|
| Measurement accuracy | 0.08 mm | Variable, operator dependent | Consistent and repeatable |
| Detection time per flange | 1.2 s | 4 to 5 s | 70% faster |
| False detection rate | Low | High | Significantly reduced |
| Data recording | Automatic | Manual | Eliminates transcription errors |
| Operator skill requirement | Minimal | High | Reduces training burden |
The sub-pixel edge detection capability of the Canny operator is a key factor in achieving the 0.08 mm accuracy. Sub-pixel methods estimate the edge position to a fraction of a pixel, which translates to a measurement resolution significantly better than the camera's native pixel size. The least squares fitting further improves accuracy by averaging the edge points and reducing the impact of individual measurement errors.
Engineering Practice Considerations
For automotive pipe fitting manufacturing, flange dimension accuracy is critical for ensuring proper assembly and sealing. The flange face dimensions, including diameter, thickness, and concentricity, directly affect the fit with mating components and the sealing performance. Key practical considerations include:
- Lighting conditions: The lighting must be uniform and controlled to avoid shadows and reflections that can affect edge detection. Ring lighting or structured lighting is typically used for flange inspection.
- Workpiece positioning: The flange must be positioned consistently on the inspection stage to ensure that the camera captures the correct view. Fixturing or alignment aids may be required.
- Surface quality: Surface defects such as scratches, burrs, or oxide scale can interfere with edge detection and must be accounted for in the image processing algorithm.
- Calibration maintenance: The camera calibration must be verified periodically to ensure that the measurement accuracy is maintained. Environmental factors such as temperature changes and mechanical vibration can affect calibration.
- Integration with production line: The system should be integrated with the production line's control system to enable real-time feedback and quality control.
Study Insights and Reflections
The machine vision approach to flange dimension detection represents a significant advancement over manual inspection. The combination of sub-pixel edge detection and least squares fitting provides a robust method for achieving high measurement accuracy without the need for expensive coordinate measuring machines. The 1.2 second detection time is fast enough for online inspection, allowing for real-time quality control and immediate rejection of non-conforming parts.
One important insight is that the system's accuracy of 0.08 mm is sufficient for most automotive pipe fitting applications, where the typical dimensional tolerance for flange faces is plus or minus 0.1 to 0.2 mm. However, for high-precision applications such as fuel system components or brake system fittings, tighter tolerances may be required, and the system may need to be upgraded with higher-resolution cameras or more sophisticated image processing algorithms.
Another consideration is the robustness of the system against variations in workpiece appearance. Different batches of pipe fittings may have different surface finishes, colors, or coatings, which can affect the image processing results. The system should be tested with a representative range of workpiece conditions to ensure reliable performance across production variations.
The integration of HALCON and C# provides a practical and extensible software architecture. HALCON offers powerful image processing libraries, while C# provides a familiar programming environment for developing the user interface and integrating with production systems. This combination is well-suited for industrial applications where reliability and maintainability are important.
Summary
This paper presents a practical and effective machine vision-based dimension detection system for automotive pipe fitting flanges, achieving a measurement accuracy of 0.08 mm and a detection time of 1.2 seconds per flange. The system combines HALCON image processing with C# programming to provide a robust solution for online quality control. For engineering practice, the key takeaway is that machine vision technology can significantly improve the efficiency and consistency of flange dimension inspection, reducing reliance on manual inspection and enabling real-time quality control in production environments.