Point cloud simplification with preserved edge based on normal vector

This paper presents a point cloud simplification algorithm with preserved edge based on normal vector. Because edge points have more distinct features than non-edge points, these special points should always be preserved in the point cloud simplification process. The proposed algorithm establishes the spatial topology relationship for each point using octree first, then identifies and retains edge points using a simple but effective method. For non-edge points, delete the least important points until user-specified data reduction ratio is reached. The importance of a non-edge point is measured using the average of the Euclidean distances (based on normal vector) from the point to estimated tangent plane at its each neighborhood point. The experimental results on three test point cloud data sets and two practical data sets of our own demonstrate that the proposed algorithm performs much better compared with other methods.