影像量測技術已廣泛應用於結構工程與地震工程中，相較於傳統的量測方法，影像量測不會因工作空間導致架設困難。在過去的研究中，前人已成功將影像量測技術應用於油壓致動器即時位移回饋控制中，發展出一套非接觸式的位移回饋控制方法。由於影像量測計算之更新頻率遠低於油壓致動器之更新頻率，導致油壓致動器階梯狀訊號的產生；此外，在追蹤靜止不動或慢速移動的目標時，由於重複精度導致計算之影像位移訊號產生上下跳動，造成致動器位移控制表現產生微小的振動。前人已透過加入邊緣搜索以及次像素精度解決重複精度問題，更透過加入中介層將影像量測回饋的更新速率提高10倍。由於過去的研究僅使用單向的位移回饋控制，本研究嘗試將此技術延伸應用在雙向的位移回饋控制實驗中。
本研究首先透過將兩台相機以正交架設的方式，建構成一個正交單相機系統，並配合低差異採樣法來算出目標物的位移，以一台相機所計算出位移對應一支致動器的方式，完成了非接觸式的雙向位移回饋控制。然而，單相機系統需要確保相機與試體呈幾何正交，在結構實驗配置與實驗空間限制下通常不太可行。有鑑於此，本研究提出了以雙相機系統進行影像量測位移回饋控制的概念，稱為雙目視覺系統。首先，即時搜尋兩台相機所得影像中目標物的影像點，以三角測量法的方式找出真實點的位置，再計算出試體在兩個方向的真實位移，以此來進行致動器雙向位移回饋控制。最後，本研究於國家地震工程研究中心小型結構實驗室進行單相機與雙相機系統的實驗驗證，實驗結果顯示兩種影像量測雙向致動器位移回饋控制技術皆可得到良好的位移控制性能，同時影像量測的更新頻率亦不因計算量變大而下降。

Nowadays, image-based analysis has been widely applied to structural engineering and earthquake engineering studies as noncontact measurements. Compared to conventional displacement transducers, the location for setting up cameras is not limited by experimental layout and working space. In previous studies, image-based analysis technology has been successfully applied for real-time displacement feedback control of actuator to replace conventional displacement transducers. However, the insufficient update frequency of image analysis resulted in step-like command for actuator control. Meanwhile, displacement jitter occurred when the loading rate was slow due to the repeatability of the camera and the analysis. Accordingly, the displacement tracking performance of the hydraulic actuator is affected. Later, the repeatability issue of the camera and analysis has been resolved by incorporating edge detection and sub-pixel accuracy algorithm. The step-like signal has also been improved by increasing the update rate tenfold through the a middleware which generates virtual displacement feedbacks by extrapolating the displacement measurement from image analysis. These previous researches merely focused on unidirectional displacement feedback control of actuator. This study intends to extend the image analysis technology to bidirectional displacement feedback control of actuators.
Two scenarios are considered in this study. First, a single-camera system is defined by setting up two cameras orthogonal to their corresponding actuators. The displacement of the target object is calculated using the same technology developed for unidirectional control of actuator which employs low-discrepancy sampling method. The displacement calculated by one camera corresponds to the movement of the associated actuator. In other words, the two cameras are working individually without recognizing each other. However, it may not be always allowed to set up two cameras in orthogonal directions due to complicated experimental layout or a lack of space in the laboratory. Therefore, a second scenario considers using a stereo vision system for bidirectional displacement feedback control of actuators. The image points of the target object are obtained by the two cameras simultaneously. The position of the true points are then determined using the triangulation method. Afterwards, the actual displacement of the test specimen in both orthogonal directions can be calculated which is used as feedback for bidirectional displacement control of actuators. To verify the feasibility of the proposed method, experiments are conducted in the small-scale structural laboratory of National Center for Research on Earthquake Engineering. The experimental results demonstrate that applying image measurement techniques for bidirectional displacement feedback control achieves satisfactory static and dynamic displacement tracking performance without reducing the update frequency of image analysis.

1. 鄭宇辰, 應用影像量測技術於致動器位移回饋控制之研究. 2020.
2. 龔柏亘, 即時影像量測技術於致動器位移回饋控制性能提升與應用之研究. 2022.
3. Benhimane, S. and E. Malis. Real-time image-based tracking of planes using efficient second-order minimization. in 2004 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE Cat. No.04CH37566). 2004.
4. Malis, E. Improving vision-based control using efficient second-order minimization techniques. in IEEE International Conference on Robotics and Automation, 2004. Proceedings. ICRA '04. 2004. 2004.
5. Wang, K., et al. A pattern matching method using geometric information of images. in 2011 6th IEEE International Conference on Nano/Micro Engineered and Molecular Systems. 2011.
6. Yang, Y.S., Wu, C. L., Loh, C. H., Yang, C. M., Tu, W. H., Shih, M. H., and Tung, S. H., Preliminary Study on Image-based Measurement. 2008. 13: p. 1-47.
7. Yang, Y.S., Huang, C. W., and Wu, C. L., A simple image-based strain measurement method for measuring the strain fields in an RC-wall experiment. Earthquake Engineering & Structural Dynamics, 2012. 41(1): p. 1-17.
8. Chen, P.-C., Y.-S. Yang, and Y.-C. Cheng, Displacement feedback control of actuators for structural testing using image processing and analysis. Earthquake Engineering & Structural Dynamics, 2022. 51(3): p. 630-647.
9. Lin, X., J. Wang, and C. Lin. Research on 3D Reconstruction in Binocular Stereo Vision Based on Feature Point Matching Method. in 2020 IEEE 3rd International Conference on Information Systems and Computer Aided Education (ICISCAE). 2020.
10. Li, T.-Y.a.Z., H. -B. Research on model control of binocular robot vision system. in 2018 Chinese Automation Congress (CAC). 2018.
11. Zhao, F. and Z. Jiang. A New Algorithm for Three-dimensional Construction Based on the Robot Binocular Stereo Vision System. in 2012 4th International Conference on Intelligent Human-Machine Systems and Cybernetics. 2012.
12. Xi, L., et al. A method of stereo images rectification and its application in stereo vision measurement. in 2010 Second IITA International Conference on Geoscience and Remote Sensing. 2010.
13. Li, Z. and Z. Jianping. A Step-step Calibration Method Based on Two-eye Stereo Vision Measuring Robot. in 2007 8th International Conference on Electronic Measurement and Instruments. 2007.
14. Instruments, N., IMAQ Vision Concepts Manual. p. 1-313.