Algorithm 2.

Obstacle detection based on connected component labeling approach

	Input:
	G: horizontal or vertical occupancy map
	b: background value
	Output:
	obL: obstacle bounding box (position, size: [px, py, sx, sy]) list
1:	procedure Obstacle Detection(G,b)
2:	L ← G	▹ initialization: Label
3:	(W,H) ← L.size()	▹ initialization: Width, Height
4:	childList, labelSet, label ← empty
5:	for w, h in W, H do	▹ label it
6:	if L(w, h) ! = b then
7:	update label
8:	L(w, h) ← label
9:	append L(w, h) to childList if its top/left exists
10:	for w, h in W, H do	▹ label updating
11:	if L(w, h) ! = b then
12:	if L(w, h) in childList then
13:	update it as its father
14:	else
15:	add it in labelSet
16:	for label in labelSet do	▹ get result
17:	ob ← bounding_box(label)
18:	append ob to obL
19:	return obL