Research On Workpiece 3D Reconstruction Technology Based On Multi-View

Multi-view 3D reconstruction based technology is used in a wide range of fields because of its low cost,universality and realistic texture colours.In recent years,as industrialisation continues to develop,there has been a great deal of interest in the research of 3D reconstruction techniques for workpieces.The Monocular multi-view 3D reconstruction technology simplifies the requirements for equipment,scenes and personnel skills,and can complete the 3D reconstruction task with only one high-quality imaging camera.However,the current Monocular multi-view 3D reconstruction technology still has the problem of difficult extraction of feature points and low accuracy of feature alignment for some workpieces with weak texture areas.Therefore,Aiming at this,this paper proposes a Surface Mesh refinement Algorithm based on Shape From Shading(SFS)to address the problem that texture detail is smoothed or not recovered in the reconstruction of texture-rich artefacts.The algorithm takes the Delaunay mesh model as the initial value.The mesh vertices are then projected onto the visual image to obtain their corresponding SFS depth values.Furthermore,the photometric consistency constraint algorithm is used to fuse the SFS constraint algorithm as the data term and the model’s own curvature constraint as the smoothing term,which together form the energy function for the mesh vertex optimisation,and finally the mesh vertices are adjusted by minimising the energy through gradient descent to optimise the mesh model.The experiments show that the mesh model optimisation algorithm incorporating SFS in the DTU dataset improves the accuracy of the input initial mesh model by 14.5% and the accuracy of the existing mainstream surface mesh model optimisation algorithm by 12%.At the same time,the mesh model optimisation algorithm incorporating SFS in the self-built part dataset improves the accuracy of the input initial mesh model by 8% and the accuracy of the existing mainstream surface mesh model optimisation algorithm by 2.5%.At the same time,in order to improve the workpiece reconstruction accuracy.A light source calibration algorithm is added to optimize the depth of SFS.Control experiments were added to prove that the accuracy of mesh model optimization algorithm based on fusion SFS with additional light source calibration was increased by 44.5% compared with that without additional light source calibration.Therefore,it can be seen from the experimental results that the mesh optimization algorithm integrated with SFS has a better effect than the existing open source mesh model optimization algorithm.At the same time,if the location information of light source is added,the accuracy of grid model is higher.Therefore,the algorithm can effectively correct the texture details of the surface mesh model,improve the precision of the mesh model,and improve the effect of the reconstruction model.