TL;DR
This paper introduces an automated camera exposure control system for stereo-digital image correlation (stereo-DIC) that ensures high-quality image capture under varying ambient light. By automatically adjusting camera exposure, the system achieves more accurate 3D deformation measurements, particularly in challenging environments like high-temperature tests.
What Problem Does This Paper Address?
Stereo-digital image correlation (stereo-DIC) is a powerful technique for measuring full-field 3D shape and deformation. However, its accuracy heavily relies on the quality of the captured speckle images. Traditionally, achieving optimal image quality requires manual adjustment of camera exposure, aperture, and lighting. This process is time-consuming, subjective, and prone to errors, especially when ambient light conditions change during a test (e.g., due to thermal radiation in high-temperature experiments). Manual adjustments can also inadvertently alter system calibration parameters. Existing automated methods for 2D-DIC struggle to directly translate to stereo-DIC due to the need for synchronized image capture from two cameras with identical exposure times and the inherent intensity differences between left and right images, which can degrade stereo matching accuracy. This paper addresses the need for an automated, robust, and accurate exposure control method specifically for stereo-DIC systems.
Methods
The proposed method builds upon previous work in automated exposure control for 2D-DIC but adapts it for stereo-DIC. It operates in two main stages:
- High-Quality Reference Image Pair Acquisition:
- The system first synchronizes the two cameras and adjusts illumination and lens apertures to ensure the average grayscale (AG) values of the left and right images are similar.
- Individually, each camera's exposure time is adjusted to find the optimal exposure that maximizes the Mean Intensity Gradient (MIG), a metric for speckle pattern quality. The corresponding optimal AG values for each camera are also determined.
- An "ideal average grayscale" (gopt) is calculated as the average of the two optimal AG values.
- Finally, the exposure time for both cameras is adjusted to achieve this ideal average grayscale, ensuring a high-quality reference image pair with minimal intensity difference between the left and right images.
- High-Quality Deformed Image Pair Acquisition:
- During testing, the system adaptively adjusts the camera exposure time to maintain the AG of the left image close to the predetermined ideal value (gopt). This is achieved through an iterative process where the exposure time is updated based on the current AG value and the target ideal AG value.
- A confidence region is defined (±10% of gopt) to determine if the image quality is acceptable. If within this region, the current exposure time is used. If outside, the exposure time is iteratively adjusted until the condition is met. This adaptive adjustment allows the system to compensate for varying ambient light conditions without manual intervention.
Datasets
The paper does not explicitly mention or introduce new datasets for training. Instead, it focuses on experimental validation using a custom-built stereo-DIC system. The experiments involve capturing image pairs under various controlled conditions, including:
- Varying illumination levels (from 50 to 250 levels).
- Translation tests (in-plane and out-of-plane) of a planar sample.
- High-temperature experiments on a nickel-based alloy specimen up to 800 °C.
These experimental setups and the resulting image data serve as the "datasets" for evaluating the proposed method's performance against a fixed exposure mode.
Key Insights and Ideas
- Bridging the Gap for Stereo-DIC: The core innovation is adapting an automated exposure control concept from 2D-DIC to the more complex stereo-DIC setup, specifically addressing the challenges of synchronized capture and inter-camera intensity differences.
- MIG as a Quality Metric: The reliance on Mean Intensity Gradient (MIG) as a key indicator of speckle pattern quality is crucial. Maximizing MIG ensures sharp, well-defined speckles for accurate correlation.
- Adaptive Exposure for Dynamic Environments: The adaptive adjustment of exposure time during testing is vital for maintaining image quality in environments with fluctuating ambient light, such as high-temperature tests where thermal radiation can significantly alter lighting.
- Ideal Average Grayscale (gopt): Defining a target average grayscale value (gopt) that balances the optimal conditions for both cameras is a clever way to achieve consistent image quality across the stereo pair.
- Robustness to Ambient Light: The experimental results demonstrate that the automated system maintains consistent image quality (MIG and AG values) even when ambient light intensity changes drastically, outperforming fixed exposure methods that lead to over-saturation or under-exposure.
Key Results
- Improved Image Quality: The proposed automated exposure control method consistently captures high-quality reference and deformed image pairs with better contrast and higher MIG values compared to fixed exposure methods. For example, in validation tests, MIG values for the optimal reference image pair were higher than those from a fixed mode comparison.
- Enhanced Measurement Accuracy:
- Temporal Matching: The auto exposure mode yielded significantly better temporal matching, evidenced by higher average ZNCC coefficients (close to 1.00 with low standard deviation) compared to the fixed exposure mode, where SD increased with light intensity.
- Stereo Matching: Stereo matching results were also superior with the auto exposure mode.
- Displacement Errors: In translation tests, the mean errors and standard deviations for U (in-plane) and W (out-of-plane) displacements were less than half those obtained with the fixed exposure mode.
- Robustness: The system demonstrated robustness against varying ambient light. Even at high light intensities (250 level), images captured with auto exposure showed good contrast, while fixed exposure resulted in over-saturation.
- High-Temperature Performance: In real high-temperature experiments (up to 800 °C), the auto exposure mode successfully captured usable images, allowing for the retrieval of full-field thermal deformation and accurate estimation of the coefficient of thermal expansion (CTE), whereas the fixed exposure mode failed due to over-saturation.
Limitations and Future Work
- Limitations:
- The paper acknowledges that while the method aims to minimize intensity differences, some inherent differences between left and right images may still exist due to camera settings and viewing angles, which can still slightly impact stereo matching.
- The adaptive adjustment relies on the AG value of the left image to control exposure for both cameras. While effective, this assumes a correlated response between the cameras to ambient light changes.
- Future Work:
- The authors suggest that the proposed method is highly recommended for practical use in various stereo-DIC systems to ensure optimized deformation measurement.
- While not explicitly stated as future work, the inherent limitations point towards potential areas for improvement, such as more sophisticated algorithms to further minimize inter-camera intensity differences or exploring alternative quality metrics beyond MIG and AG. The paper's success in high-temperature tests also opens avenues for applying this method to other extreme environments.
Summary updated