Two-Level Model for Detecting Substation Defects from Infrared Images

. 2022 Sep 10;22(18):6861. doi: 10.3390/s22186861

Algorithm 1 TDDM

1:
Input: Infrared image c, Number of superpixels K.
2:
Output: All defect regions in the infrared image.
3:
Obtain instance $c_{m} = Seg (c), m = 1, 2, \dots, M$
4:
for $m = 1 to M$ do
5:
for $n = 1 to N$ do
6:
Compute superpixels $c_{m}^{n}$
7:
Obtain temperature characteristic $T_{m}^{n}$
8:
end for
9:
Compute temperature probability density distribution $f_{m}$
10:
Compute the local maximum $O_{m}^{q} (x_{m}^{q}, y_{m}^{q})$ of $f_{m}$ , where $q = 1, 2, \dots, Q$
11:
if $Q \geq 3$ then
12:
for $n = 1 to N$ do
13:
if $T_{m}^{n} > x_{m}^{3}$ then
14:
Determine $c_{m}^{n}$ defective
15:
else
16:
Determine $c_{m}^{n}$ normal
17:
end if
18:
end for
19:
Merge all adjacent defective superpixels to obtain $D_{m}$
20:
else
21:
Output: No defect in the instance.
22:
end if
23:
end for