A Robust 3-D Abrasion Diagnosis Method of Pantograph Slipper Based on Stereo Vision

Dong Li,Zhen Liu,Shengfang Lu,Luonan Chang

DOI: https://doi.org/10.1109/tim.2020.2995486

IF: 5.6

2020-11-01

IEEE Transactions on Instrumentation and Measurement

Abstract:The pantograph is an essential component in modern electrified trains. Due to the characters of contactless and high precision, the 3-D stereo vision, which can effectively monitor the abrasion, is of great significance for ensuring train safety. However, illumination conditions, such as overbrightness and insufficient light, have a great impact on semiglobal matching (SGM) stereo matching, inevitably affecting the integrity and accuracy of the abrasion measurement of the pantograph slipper. In this regard, a robust SGM stereo matching is proposed in this article for the 3-D abrasion measurement. First, the combination of the neighborhood information-aware census cost and multiscale fusion for cost aggregation is put forward. While insensitive to illumination condition changes, the improved census cost can quantify the texture information of the neighborhood. Moreover, multiscale cost aggregation further fuses the neighborhood information of different scales. As a result, the performance of SGM in weakly textured areas is greatly improved, especially in complex illumination conditions. Eventually, with the help of the subpixel disparity refinement and local points fitting, an accurate 3-D abrasion calculation is achieved. Practical experiments show the robustness of stereo matching. The precision verification and computation time analysis confirm the measurement accuracy and applicability.

engineering, electrical & electronic,instruments & instrumentation

What problem does this paper attempt to address?

The paper primarily focuses on addressing the issue of pantograph slide wear detection in modern electrified trains. Specifically, the paper attempts to solve the following key problems:

1. **The need for automatic wear detection**: The current manual inspection methods are inefficient and labor-intensive, thus necessitating an automated wear diagnosis method as a replacement.

2. **Limitations of existing methods**:
- Image processing-based methods (e.g., [5]-[9]) can only provide qualitative predictions and rely on strict indoor lighting conditions.
- Methods using 3D laser sensors (e.g., [10]-[13]) are limited by sensor accuracy and the requirement for the pantograph to be strictly parallel to the laser plane.

3. **Challenges under complex lighting conditions**: Existing methods struggle to adapt to complex lighting conditions during field measurements, such as uneven illumination, overexposure, or low brightness. These conditions severely affect image quality, increase edge feature sensitivity, and consequently impact detection accuracy.

4. **Improving stereo matching algorithm performance**: The paper proposes an improved semi-global matching (SGM) method to address the stripe hole problem in stereo matching under complex lighting conditions. By introducing improved neighborhood information-aware census cost and multi-scale fusion, the matching performance in weak texture areas is enhanced.

5. **Achieving high-precision 3D wear measurement**: Combining sub-pixel disparity refinement and local point fitting methods, the paper proposes a new method for 3D wear measurement, capable of performing wear assessment with high precision and efficiency.

In summary, the paper aims to develop a novel, robust 3D wear diagnosis method based on stereo vision to meet the needs of actual field measurements and to address the limitations of existing technologies under complex lighting conditions.