Machine Vision-based Optical Waveguide Array Fiber Alignment Technology

Offering the advantages of compactness, immunity to electromagnetic interference, performance consistency, excellent stability and reliability as well as suitability for automated production, et al, planar optical waveguide devices have been the leading research concern in the optoelectronic field. Up to now, packaging is becoming the bottleneck for its further development and extensive applications. Meanwhile, the alignment technique is the key to the packaging process.Considering the sensitivity to the fabrication errors, the difficulty and high cost to meet the requirements in current methods using fixtures and motion stages with high positioning accuracy and repeatability, this paper provides a packaging process of which the coarse alignment is a combination of using machine vision and raster-scanning method.For a machine vision system, image acquisition is the basis of its task fulfillment. In order to observe the key features of the fiber array and waveguide chip precisely, a vision system with two orthogonal cameras is presented and the selections of cameras, lens along with illumination scheme are accomplished elaborately.A widely used camera model is described in detail. Based on a comparative review of existing camera calibration methods, a time-saving and effective calibration approach is designed to solve the pixel-metric transformation. It ensures the stability and accuracy of the vision system.A new straight line fitting based on two points is presented in this paper. It is demonstrated that the new method is relatively more accurate in result,time-saving in process as well as antinoise, compared with Hough and LSM algorithm. The spacing and angle between fiber array and waveguide chip are calculated precisely based on the straight line segment features and then the elementary alignment is carried out.