The DSO Feature Based Point Cloud Simplification

This study proposes an effective low-error point cloud simplification method to retain the physical features of the models. The value of Discrete Shape Operator (DSO) is adopted to extract the features points of the models, and those are postponed to simplify. The value of DSO is defined as the discrete sum over the directional curvature and torsion. The proposed method improves the Quadric Error Metric of vertex pair contraction, it not only effectively simplifies the point cloud model and keeps the features of object model, but also decreases the preprocessing time cost associated with feature analysis. This study also proposes a method to obtain unique simplified model for each model and the time cost involved in calculating DSO is about 17.29% of the execution time. The unique simplified model obtained by this study can significantly reduce the computation cost about 72.72% than mesh simplification which reconstruct original points first.