Sharp Feature Reconstruction Method for Triangular Meshes in Additive Repair
In the process of additive repair,the point cloud data obtained by scanning the workpiece with structured light is usually encapsulated into a triangular mesh.Due to interference in the scanning environment or blurring of sharp features during point cloud post-processing,the sharp features that originally existed in the workpiece will be lost,making it impossible to restore the true appearance of the workpiece in the final triangular mesh.To solve this problem,this paper designs an algorithm for reconstructing sharp features in triangular meshes and studies the detection and reconstruction techniques of sharp features in triangular meshes.Firstly,the angle between the normal vector direction of each triangular patch and the normal vector direction of its neighboring faces is detected.The selected neighboring face is a ring of neighboring faces of the triangular patch,that is,the triangular patch has common edges or vertices with these neighboring faces.If there is an angle exceeding a certain threshold between the normal vector of the patch and the normal vector of the neighboring face,the triangular patch is set as a feature patch,and the feature patch is recorded and saved in a set.Next,traverse each feature patch in the set,determine whether it is the outermost feature patch,that is,whether it is adjacent to a non feature surface.If it is adjacent to a non feature surface,enter the processing step for that feature patch,find its similar patch in the non feature patches in its neighborhood,and adjust the normal vector direction of this feature patch according to the normal vector direction of the similar patch,so that the normal vector direction of this feature patch is as parallel as possible to the normal vector direction of the similar patch.Then,continue to adjust the positions of the three vertices of the feature patch,adjust the vertex positions of the feature patch according to the direction of the normal vector adjusted in the previous step,so that the adjusted triangular patch is as perpendicular as possible to the adjusted normal vector.Before adjusting each vertex,it is necessary to determine whether the vertex also belongs to a non feature surface.If the vertex also belongs to a non feature surface,the vertex position will not be adjusted.After all three vertex positions of the feature patch are adjusted,continue to traverse other unprocessed feature patches in the set.Finally,set the feature patches that have undergone the above processing as non feature patches,and repeat all the above steps until the normal vector directions and vertex positions of all feature patches are processed.All the above steps can be repeated multiple times,and after a certain number of iterations,the Laplacian denoising step can be performed to achieve better sharpening effect.The experimental results show that the running time of sharpening increases with the increase of the number of facets,and is also related to the shape of the triangular mesh model.The more feature facets there are,the longer the processing time is.The proposed sharpening method has a sharpening time of less than 10 s for processing 100 000 triangular facets.The proposed sharpening method has a relatively stable effect on triangular meshes with different densities,but the sharpening effect on overly sparse triangular meshes is poor.The sharpening effect on triangular meshes is not entirely determined by the complexity of the model's shape,but mainly depends on the sparsity degree of the triangular meshes.The proposed sharpening method can effectively sharpen triangular meshes with a large amount of noise and basically meets the requirement of reconstructing sharp features that have already been lost.
NETL
NSTL
万方数据
