A Preliminary Study of Deep Learning Sensor Fusion for Pedestrian Detection

. 2023 Apr 21;23(8):4167. doi: 10.3390/s23084167

Algorithm 1 Lidar point coordinates $x_{l}, y_{l}, z_{l}$
1:	procedure
2:	•Get the $u, v$ undirstorted point in pixel coordinates.
3:	function UndirstortPoint( $x, y$ )
4:	$u_{u} \leftarrow x$
5:	$v_{u} \leftarrow y$
6:	return $u_{u}, v_{u}$
7:	end function

8:	•Get the directional vector in the camera frame.
9:	function DirectionalVector( $u_{u}, v_{u}$ )
10:	$u_{d} \leftarrow u_{u}$
11:	$v_{d} \leftarrow v_{u}$
12:	$w_{d} \leftarrow 1$
13:	return $u_{d}, v_{d}, w_{d}$
14:	end function

15:	•Get the directional vector in the lidar frame.
16:	function Camera2LidaRotation( $u_{d}, v_{d}, w_{d}$ )
17:	$[\begin{matrix} u_{l} \\ v_{l} \\ w_{l} \end{matrix}] \leftarrow \begin{matrix} _{c}^{l} R \end{matrix} {[\begin{matrix} u_{d} & v_{d} & w_{d} \end{matrix}]}^{T}$
18:	return $u_{l}, v_{l}, w_{l}$
19:	end function

20:	•Compute the intersection point between a line and a plane.
21:	function PlaneLineCrossingPoint( $u_{l}, v_{l}, w_{l} P C_{l}$ )
22:	$\vec{u} \leftarrow [u_{l}, v_{l}, w_{l}]$	▷ creates the $\vec{u}$ vector
23:	$A x + B y + C z + D = 0 \leftarrow P C_{l}$	▷ board’s parallel plane model
24:	$\vec{n} \leftarrow [A, B, C]$	▷ plane normal vector
25:	$v_{o} \leftarrow [A, B, C, D]$	▷ a point in the model plane
26:	$p_{o} \leftarrow_{c}^{l} t$	▷ translation vector from camera to lidar
27:	$w \leftarrow p_{o} - v_{o}$
28:	$D \leftarrow \vec{n} \cdot \vec{u}$	▷ dot product
29:	$N \leftarrow - \vec{n} \cdot \vec{w}$	▷ dot product
30:	$s l \leftarrow \frac{N}{D}$
31:	$\vec{p} \leftarrow p_{o} + s l \times \vec{u}$	▷ intersection point
32:	return $\vec{p}$
33:	end function
34:	end procedure