Research On 3D Imaging And Measurement Technology For Front-end Data Acquisition Of Intelligent Manufacturing

At the front end of the intelligent manufacturing line,it is the foundation of the subsequent automatic operation of intelligent manufacturing to obtain the contour size parameters of the blank to be processed or the workpiece to be assembled and convert them to the reference coordinate system of the production line.Due to low efficiency,the traditional manual measurement and contact measurement methods cannot meet the requirements of fast,flexible and stable intelligent manufacturing.Based on the active light binocular vision measurement method,design multi-line laser structure light binocular vision measurement system,aiming at the problem of 3D contour reconstruction and dimension measurement of parts without obvious texture features,the research focuses on the extraction and matching technology of multi-line laser fringe center,scanning point cloud coarse registration and fine registration technology,and surface reconstruction technology of point cloud data.First of all,the composition of the front-end data acquisition system for industrial intelligent manufacturing is discussed,and the coordinate transformation relationship between each component is analyzed and constructed,and the coordinate system of each part is converted to the same coordinate system,so as to improve the front-end data acquisition efficiency.Then,the camera calibration of the multi-line laser structured binocular vision 3D measurement system is carried out.Aiming at the problems such as low precision and easy ambiguity matching of multi-line laser fringe center in active light source,an algorithm based on threshold segmentation and Hessian matrix was proposed to improve the extraction speed.According to the binocular vision model and camera calibration results,the stereo matching algorithm based on the polar constraint and sequence consistency constraint was used to carry out stereo matching,and the 3D coordinates of laser center line were obtained.On this basis,optical strip point cloud coarse registration is carried out based on non-coding circular marking points,aiming at the characteristics of sparse laser point cloud and few spatial geometric features.An algorithm based on the properties of elliptic polar triangle is proposed to extract the mark points on the surface of workpiece.According to the coarse registration results,the improved ICP algorithm is used to complete the fine registration of multi-point cloud,and the triangular meshing algorithm is used to realize the surface reconstruction of point cloud data.A software and hardware platform for 3d measurement system of multi-line structured light binocular vision is designed and built.3d reconstruction and measurement experiments are designed for workpiece with different surface characteristics.The experimental results show that the method proposed in this paper has a good effect of surface contour reconstruction for both curved surface parts and stepped plane parts.The measurement accuracy can be up to 0.2mm within the measurement range of 80cm～120cm.It can meet the positioning requirements of automatic cutting of blank parts in intelligent manufacturing production line,or the positioning requirements of automatic grasping of assembly parts by manipulator.