Algorithm 1. CSNDT
Input:
$X$: Current scan
$Y$: Reference scan
Output:
$p$: Transform parameter
1:	{Initialization:}
2:	$Y=\{C_1, C_2, \dots , C_n\}$ ← L-DBSCAN
3:	For all Point cloud cluster $C_i\in Y$ do
4:	$C_i=\{m_1,\dots ,m_m\} \leftarrow$ all points in $C_i$
5:	${\mu }_j\leftarrow {\displaystyle \frac{1}{n}}{\sum }_{i=1}^m{m}_{\mathrm{j}}$ $m_{\mathrm{j}}\in C_i$
6:	${\sum }_j\leftarrow {\displaystyle \frac{1}{n}}{\sum }_{j=1}^m(m_j-{\mu }_j){(m_j-{\mu }_j)}^T$
7:	$l=\max (x_{\max }-x_{\min }, {\mathrm{y}}_{\max }-y_{\min })$
8:	If $seg>1$ do segmentation
9:	For all small Point cloud cluster $C_i^s\in C_i$ do
10:	$C_i^s=\{m_1,\dots ,m_m\} \leftarrow$ all points in $C_i^s$
11:	${\mu }_j^s\leftarrow {\displaystyle \frac{1}{n}}{\sum }_{i=1}^m{m}_{\mathrm{j}}$ $m_{\mathrm{j}}\in C_i$
12:	${\sum }_j^s\leftarrow {\displaystyle \frac{1}{n}}{\sum }_{j=1}^m(m_j-{\mu }_j^s){(m_j-{\mu }_j^s)}^T$
13:	End for
14:	End if
15:	End for
16:	{First Registration:}
17:	While not converged do
18:	$score\leftarrow 0, g\leftarrow 0, \mathrm{H}\leftarrow 0, p_1\leftarrow 0$
19:	For all points $x_i\in X$ do
20:	Find the cell $F_i$ that contains $x_i^{\prime }$
21:	$score\leftarrow score+p(x_i^{\prime })$
22:	Update $g, H$
23:	End for
24:	Solve $H\Delta p_1=-g$
25:	$p_1\leftarrow p_1+\Delta p_1$
26:	End while
27:	{Second Registration:}
28:	While not converged do
29:	$score\leftarrow 0, g\leftarrow 0, H\leftarrow 0, p_2\leftarrow p_1$
30:	For all points $x_i\in X$ do
31:	Find the cell $F_i^s$ that contains $x_i^{\prime }$
32:	$score\leftarrow score+p(x_i^{\prime })$
33:	Update $g, H$
34:	End for
35:	Solve $H\Delta p_2=-g$
36:	$p_2\leftarrow p_2+\Delta p_2$
37:	End while