A New Simplification Algorithm for Point Cloud Based on the Vertical Plane Constraint and Moving Window

Abstract

Using three-dimensional spatial information, this method constructs the detection condition, compares the conditions to be detected, formed by the points in the file to be simplified, then determines the redundancy of the points. Moving windows are used to promote the operation of the algorithm and generate many tiny approximate vertical planes, from which the simplified points will be generated. By selecting experimental data on rabbit and horse and comparing the methods based on mesh and curvature, the proposed method increased the simplification rate of rabbit and horse from 3.022% and 11.123% (mesh method), and 5.704% and 15.316% (curvature method), to 83.387% and 84.296%, respectively, while the standard deviation was reduced from 0.01051 and 0.0157 (mesh method), and 0.0817 and 0.0013 (curvature method), to 0.02179 and 0.01507, respectively. For the simplification of multiple objects, the proposed method increased the simplification rate from 89.113% and 91.826% (mesh method), and 84.79% and 88.91% (curvature method), to 93.458% and 96.916%, respectively, reducing the time by approximately 20 s.

Keywords

• Vertical plane constraint
• point cloud simplification
• normal vector angle entropy
• moving window
• feature points