Optimized Multi-Spectral Filter Array Based Imaging of Natural Scenes

. 2018 Apr 12;18(4):1172. doi: 10.3390/s18041172

Algorithm 1 The Proposed DPSO Method

1:
Input: Set of candidate channels: ${M_{j} (λ) | j = 1, \dots, m}$ ; Number of output channels: n; Pattern $P$ .
2:
Output: Selected ordered set of n channels; Decided demosaicing order $O$ .
3:
Initialization:
4:
Generate population $X_{i}$ and velocity $V_{i} (1 \leq i \leq P)$
5:
$[X^{g b}, O^{g b}] = f i n d b e s t (X_{i}^{p b}, P)$
6:
for j = 1 … G do
7:
for i = 1 … P do
8:
$X_{i}^{p b} = f (M_{X_{i}}, P) < f (M_{X_{i}^{p b}}, P) ? X_{i} : X_{i}^{p b}$
9:
$V_{i} = w (V_{i - 1} + c_{1} r * (X_{i}^{p b} ⊖ X_{i}) + c_{2} r * (X^{g b} ⊖ X_{i}))$
10:
${\bar{V}}_{i} = r e d u c e m a t (V_{i}, p)$
11:
$X_{i} = X_{i} \oplus {\bar{V}}_{i}$
12:
end for
13:
$[X^{g b}, O^{g b}] = f i n d b e s t (X_{i}^{p b}, P) (1 \leq i \leq P)$
14:
end for
15:
return n ordered channels $X^{g b}$ , demosaicing order $O^{g b}$ ;