Skip to main content
. 2021 Sep 2;21(17):5909. doi: 10.3390/s21175909
Algorithm 1. The process of the real-time 3D reconstruction method
1: The RGB and Depth images of the scene are collected by the monocular camera.
2: Input RGB to YOLACT++ to obtain the segmented RGB’, 2DM.
3: Input RGB’, 2DM, Depth to VJTR to get P.
4: Set p0i=1mpi/m, then let δmax1.0 and s1.
5: for Each pi
6:  Use Formula (10) to calculate the ρi.
7: for i = 1 to m step do
8:    Use Formula (11) to calculate the δi.
9:    if (δi>δmax) then
10:      δmaxδi.
11:      si.
12:    end if
13: end for
14: end for
15: for i = 1 to m step do
16: if(pip0<λpsp0) then
17:    Add pi to Pin.
18: else
19:    Add pi to Pout.
20: end if
21: end for
22: for Each pi
23:   Use Formulas (12)–(14) to normalize its coordinate values.
24:   Add pi to P¯.
25: end for
26: return P¯.