Study On Control For Precision Alignment Of Optical Fibers Based On Telecentric Binocular Micro-vision Servoing

The alignment and fusion of the optical fiber play a fundamental role in the fiber-optic communication network. The high-precision and low-loss alignment for the optical fiber can contribute to an excellent optical fiber communication network. Because the visual servo method has many advantages such as high resolution, good robustness and non-contact measurement, up to data, it has been one of the most prevailing approaches to actively align two optical fibers. With the improvement of the manufacturing technique, the telecentric lens has already possessed numerical unparalleled advantages, such as high resolution, small distortion, and thus has great prospect development in many fields such as precision measurement. In this paper, we study the rapid alignment of the optical fibers based on the telecentric binocular micro-vision servoing.The extraction of the feature points on the optical fibers can has great influence on the control system. When the feature points are extracted more precisely, a more accuracy reconstruction of the optical fibers can be obtained and a much better alignment process can be performed. On the other hand, the frequency of the visual servoing is much higher and the alignment process can be much more stable when the speed of the image processing is faster.However, it is very difficult to improve the accuracy of the feature points while speeding the image process at the same time. Therefore, in this paper, we proposed a fast image processing algorithm based on the spatial moment with sub-pixel accuracy for the feature points.The three-dimensional reconstruction of the optical fibers calls for a high demand of the camera calibration and the feature points matching. Because of the limited FOVs and the unique imaging model of the telecentric lens, there exists the multi-solution problem while recovering the external parameters. At last, the feature points matching can also has a greet influence on the 3-D reconstruction of the optcal fibers. Thus, we introduced a fast calibration method especially for the telecentric lens and proposed a simple process to obtain the external parameters. Meanwhile, we used the fundamental matrix combined with the geometric constraint to obtain the matching points and reconstruct the optical fibers in the 3-D space.At last, we programed with the MFC language to align the optical fibers automatically.After the eye-to-hand calibration for the moving stages, the two-step alignment strategy andthe control algorithm based on PBVS(position based visual servo) were proposed and the5-DOFs alignment of the optical fibers has been fulfilled. At the same time, we analyzed the influences of various factors which can affect the process or the result of the alignment, such as the camera calibration and the speed of the stage, on the optical fiber alignment system.