Measuring small deformations of large structures using a digital camera
Deformations in large structures such as bridges, buildings, and pipelines can be dangerous. In addition to safety assessments, measuring the static and dynamic displacement of in-service structures is an important issue and a challenging task for validating designs and monitoring performance. Long-term measurement allows continuous monitoring of structures, which permits constant safety oversight so that an immediate warning can be issued if extreme displacements are detected.
Both the Global Positioning System (GPS) and laser Doppler vibrometers (LDVs) have successfully monitored the health of large structures. These instruments, however, are basically point measurement systems and are quite costly for distribution analysis. Therefore, we would prefer a monitoring system that uses remote imaging and image processing techniques to evaluate deformation in these structures. A. Mazen Wahbeh and coworkers presented a vision-based approach for measuring displacements of vibrating systems that uses a high-resolution digital camera and two high-power red LEDs.1 Similarly, Jong Jae Lee and Masanobu Shinozuka presented a vision-based system for remote sensing of bridge displacement that incorporates a camera and four white spots in a known geometry.2 Using another image processing technique, Satoru Yoneyama and colleagues applied the digital image correlation (DIC) method to measure the deflection distribution of a real bridge.3 In this method, a random pattern was used to measure the displacement distribution of the surface by comparing digital images taken before and after deformation.
We recently developed a fast, simple, and accurate small-displacement distribution measurement called the sampling moiré method.4 By analyzing digital images of a regular repeated pattern with known pitch on the structure, we can calculate deformation across the structure. Whereas the DIC method works with a random pattern, the sampling moiré method does just the opposite, reducing processing complexity by requiring the structure to incorporate a regular high-contrast grating pattern.
Figure 1 shows the fundamental principle of the sampling moiré method for single-shot phase analysis.5 Note that only one horizontal line of pixel data from the CCD image is shown here. After image processing using both the down-sampling and intensity interpolation techniques,6 multiple phase-shifted moiré fringes can be generated simultaneously. Then the phase distribution of the moiré fringe can be determined by the phase-shifting method. Finally, measuring the phase differences between the moiré fringes before and after deformation allows us to directly determine the displacement distribution.
The sampling moiré method is a promising measurement technique for research in optical methodology, such as 3D shape and strain measurements,7 because the measurement accuracy of the sampling moiré approach can reach 1/500 of the pitch of the grating.8
As a demonstration of a practical industrial application, our group showed that sub-millimeter thermal deformation can be successfully detected by the sampling moiré method.9 Figure 2 shows the displacement measured in high-temperature large-scale piping in a thermal power plant subjected to operating conditions. The target is a 1.7m-long reheat pipe, attached by two flange joints. The pipe on the left side of flange A connects to the turbine, while the pipe on the right side of flange B connects to the boiler. Under operating conditions, the temperature inside the pipe is nearly 300°C. To analyze the complete deformation behavior of the pipes, we placed 50 magnets patterned with gratings on them. The magnets were small, measuring only 100×100mm, and the grating pitch was 15mm. Using this setup allowed us to measure the displacement at each grating magnet when the temperature changed. Figure 2(a) shows the distribution of displacements at the location of each grating magnet when the temperature of the pipes decreased 8.8°C from 288 to 279.2°C. The pipes in the center moved to the left by 1.73mm, but the displacement was not constant across the pipes (note the range of colors from green to teals to blues). The displacement on the left side was smaller than on the right side by 1.11mm. Figure 2(b) shows the displacement distribution when the temperature of the pipes increased by 13.6°C (from 288 to 301.6°C): the piping moved to the right side 1.41mm at the center position, and the displacement on the right side is smaller than on the left side by 0.59mm. The direction of the displacement is the opposite of that in Figure 2(a). In other words, we successfully obtained complete displacement field information under dynamic temperature changes and measured the thermal deformation behavior of the entire pipe assembly.
In summary, a novel small-displacement measurement technique, the sampling moiré method, can monitor the full-field displacement of large structures, including high buildings and piping in power plants, as well as other large structures such as long cranes. The method requires some refinement before it can be deployed widely. We are now working to eliminate the effects of both vibration in and position change of the digital camera as well as the effects of air fluctuation between the target and the digital camera. More immediately, we are improving the measurement accuracy even while using arbitrary built-in repeated patterns (such as tile or brick walls) in large-scale structures.10 In the near future we plan to apply this method to monitor the health of large structures such as long bridges and high buildings.
Shien Ri is a senior researcher at AIST's Research Institute of Instrumentation Frontier. His research interests are optical methods, image sensing, and experimental mechanics.
