Development Of Onsite Detection And Automatic Shape Adjustment System For Formed Hull Plates Based On Line Laser Multi-camer Stereovision

At present,in the process of 3D ship hull plate bending machine working,most domestic shipyards mainly rely on experienced workers to use the wooden templates to test the formed curved plates,and follow-up shape adjustment manually through experience.The wooden templates detection method has the disadvantages of poor detection accuracy,low production efficiency and high requirements for workers' experience,which has affected the quality and speed of ship construction.In order to solve this problem,this dissertation investigates the 3D ship hull plate bending machine working site,and develops a set of onsite detection and automatic shape adjustment system for formed hull plates based on line laser multi-camera stereo vision technology.The detection system scheme design,the onsite calibration of the system,the line laser 3D reconstruction,the point cloud registration,the deviation calculation and the springback feedback of curved plates are focused in this dissertation.The ship curved plate has the characteristics of high forming accuracy and large size.Based on this,a set of onsite detection scheme based on line laser multi-camera stereo vision is proposed.The key hardware of the detection system such as industrial cameras,lenses,lasers and motion controllers are selected,and the system structure parameters are designed.Since the machining workshop of the 3D ship hull plate bending machine is susceptible to factors such as vibration and temperature difference,the structural parameters of the detection system will fail with the increase of use time.By analyzing several commonly used camera imaging models and the Zhang Zhengyou calibration algorithm based on 2D planar calibration targets,and analyzing the characteristics of the interior and exterior parameters of the system,the internal parameters of the camera and the structural parameters of the system are separated and calibrated.The specific methods are: The interior parameters of the four cameras were calibrated under offline conditions for the first time,and then the system structural parameters were calibrated periodically during the onsite procedure.This method effectively solves the problem of system structural parameters failing with time.The Gaussian filter algorithm is used to denoise the image,and the center of the laser light stripe is extracted using the extreme value method combined with the gray center of gravity method;In order to adjust the camera imaging to a completely coplanar parallel state to meet the needs of stereo matching,the Bouguet stereo correction algorithm is used to perform stereo correction on the image,which simplifies the difficulty of subsequent stereo matching;The basic principle of stereo matching in line laser binocular stereo vision is analyzed and the 3D reconstruction of ship curved plate surface is realized.This dissertation analyzes the shape adjustment file of the 3D ship hull plate bending machine,and extracts the theoretical point cloud of the curved plates through the shape adjustment file.The coarse registration of the measured point cloud and the theoretical point cloud is completed by the spatial three-point method,and the precise registration of the two point clouds is carried out by ICP algorithm;The 3D ship hull plate bending machine processes the ship curved plates with the compression force in Z direction.According to this feature,this dissertation adopts the calculation method of point cloud deviation and springback of curved plates based on projection interpolation.This method is based on the point-to-surface rule,but it does not require the construction of normal vectors and projected tangent planes,so this algorithm has the advantages of minor calculation,and the projection deviation value in the Z direction can be calculated directly,which is very suitable for the application of this dissertation.Designed and developed the onsite detection and automatic shape adjustment system function module;The standard block was measured several times.It is verified that the measurement accuracy of the detection system meets the requirements.The detection system is tested at the 3D ship hull plate bending machine processing site.The test results show that the detection system can realize the whole detection process,such as 3D measurement of the plate surface,point cloud registration,deviation calculation and feedback of the springback of the ship curved plates.