Research On Vision-Assisted High-Precision Correction Method In Laser Micro-Nano Machining

The Galvanometer system is the core equipment in Laser micro-nano processing,and its own non-linear mapping can cause distortion of the laser scanning image,which will greatly affect the processing accuracy.Therefore,distortion compensation correction of the galvanometer system is particularly important,and how to accurately extract feature points in the correction process is crucial.Traditionally,the distortion correction process of laser processing uses a method of manually measuring feature points with low efficiency and accuracy.Based on this,this thesis studies a high-precision correction method based on visual assistance,which mainly includes the following aspects:(1)Building a visual platform for Laser micro-nano processing.Researching the working principle of the laser micro and nano processing system and building the platform by adding a camera system to the laser marking machine platform.Developing a visual assistant software based on the QT framework to facilitate the design and debugging of visual algorithms,and implementing a semi-automatic calibration function.(2)Research on visual algorithms in calibration based on OpenCV methods.With the help of the developed visual assistance software,theoretical research on image preprocessing was conducted to design two grid point recognition algorithms based on the Hough algorithm and the thinning algorithm.Considering the generality of the algorithms,a grid point recognition method based on the thinning algorithm was adopted to achieve the feature point detection function of the marked pattern.To address the issue of dark corners in images caused by camera characteristics,an idea of using quadratic curve fitting was proposed to compensate and identify the grid points in the dark corner areas.(3)A correction scheme for laser scanning image distortion was designed,and a correction experiment was conducted.The scheme consisted of two parts: Camera calibration and mirror correction.Firstly,a chessboard calibration board with a Data Matrix Code was used to complete the camera calibration,and a feature point detection model for the calibration board was designed and implemented.Then,based on the study of the principle of laser scanning image distortion,a grid correction method based on bilinear interpolation was used to compensate and correct the laser spot position.(4)The research results are presented and analyzed.The camera calibration method used in this paper is compared and analyzed with the MATLAB calibration,and it is found that the calibration method based on Data Matrix Code chessboard has higher accuracy.In response to the vignetting problem,a comparative experiment is designed,and the analysis of the results shows that the method meets the high accuracy requirements.Finally,the errors after correction are evaluated,and the results satisfy the high-precision processing applications,which indirectly confirms the feasibility of the mark recognition algorithm.