Algorithm 1: Algorithm for Feature Points Extraction from Point Cloud using SCAL-SD

Input: original 3D point cloud Output: 3D Coordinates of Feature Points $\left\{{\mathit{v}}_{\mathit{C}},{\mathit{v}}_{\mathit{L}},{\mathit{v}}_{\mathit{R}}\right\}$ 1.     Group the original 3D point cloud into subsets based on pitch angle: $\left\{\phi \right\}$ 2.     For each ${\phi }_iϵ\left\{\phi \right\}$ do 3.               Apply the SCAL-SD to the points in ${\phi }_i$ to generate line clouds: $\left\{l_i\right\}$ 4.               Filter the line clouds $\left\{l_i\right\}$ to obtain the target line segments $\left\{l_1,l_2,l_3,l_4\right\}$, and derive the contour points: $P_{iL}$$, P_{iR}$ 5.     End for 6.     Fit lines to the contour points $\left\{P_{iL}\right\}$$, \left\{P_{iR}\right\}$, and determine their intersection point: ${\mathit{v}}_{\mathit{C}}$ 7.     Based on prior knowledge, determine the positions of the remaining two vertices: ${\mathit{v}}_{\mathit{R}}$$, {\mathit{v}}_{\mathit{L}}$