Robust and Fast Normal Mollification via Consistent Neighborhood Reconstruction for Unorganized Point Clouds

. 2023 Mar 20;23(6):3292. doi: 10.3390/s23063292

Algorithm 1 The pipeline of our algorithm

Input: Point cloud

P

;
Output: Normal set {

n_{i}

};
1: for

p_{i} \in P

do
2: Compute initial normal

{\tilde{n}}_{i} \in p_{i}

, feature coefficient

w_{i}

;
3: if

w_{i} < w_{t}

then
4:

n_{i} = {\tilde{n}}_{i}

;
5: else
6: Obtain mollification neighborhood

{\bar{N}}_{i} \in p_{i}

via Algorithm 2;
7: Obtain

n_{i}

via the first-stage normal mollification based on Equation (7);
8: if fitting residual

r (p_{i}) > ε_{f}

then
9: Obtain mollification neighborhood

N_{i} \in p_{i}

via Section 5.2;
10: Obtain

n_{i}

via the second-stage normal mollification based on Equation (8);
11: end for