PRS-Net: Planar Reflective Symmetry Detection Net for 3D Models

In geometry processing, symmetry is a universal type of high-level structural information 3D models and benefits many processing tasks including shape segmentation, alignment, matching, completion. Thus it an important problem to analyze various forms shapes. Planar reflective the most fundamental one. Traditional methods based on spatial sampling can be time-consuming may not able identify all planes. this article, we present novel learning framework automatically discover global planar shape. Our trains unsupervised convolutional neural network extract model features then outputs possible parameters, where input shapes are represented using voxels. We introduce dedicated distance loss along with regularization avoid generating duplicated also generalized cylinders by predicting their rotation axes. further provide method remove invalid planes demonstrate that our produce reliable accurate results. hundreds times faster than state-of-the-art methods, which sampling. robust even noisy or incomplete surfaces.