2 This section aims at justifying the lightweight attribute of our approach.
3 To be more precise, we compare the complexity of our schemes to the
4 best available steganographic scheme, namely HUGO~\cite{DBLP:conf/ih/PevnyFB10}.
7 In what follows, we consider a $n \times n$ square image.
8 First of all, HUGO starts with computing the second order SPAM Features.
9 This steps is in $O(n^2 + 2\times 343^2)$ due to the calculation
10 of the difference arrays and next of the 686 features (of size 343).
11 Next for each pixel, the distortion measure is calculated by +1/-1 modifying
12 its value and computing again the SPAM
13 features. Pixels are thus selected according to their ability to provide
14 an image whose SPAM features are close to the original ones.
15 The algorithm thus computes a distance between each feature
16 and the original ones,
17 which is at least in $O(343)$, and an overall distance between these
18 metrics, which is in $O(686)$. Computing the distance is thus in
19 $O(2\times 343^2)$ and this modification
20 is thus in $O(2\times 343^2 \times n^2)$.
21 Ranking these results may be achieved with an insertion sort, which is in
23 The overall complexity of the pixel selection is finally
24 $O(n^2 +2.343^2 + 2\times 343^2 \times n^2 + 2.n^2 \ln(n))$, \textit{i.e},
25 $O(2.n^2(343^2 + \ln(n)))$.
27 Our edge selection is based on a Canny Filter. When applied on a
28 $n \times n$ square image, the noise reduction step is in $O(5^3 n^2)$.
29 Next, let $T$ be the size of the canny mask.
30 Computing gradients is in $O(4Tn)$ since derivatives of each direction (vertical or horizontal)
32 Finally, thresholding with hysteresis is in $O(n^2)$.
33 The overall complexity is thus in $O((5^3+4T+1)n^2)$.
34 To summarize, for the embedding map construction, the complexity of Hugo is
35 dramatically larger than our scheme.
37 We are then left to express the complexity of the STC algorithm.
38 According to~\cite{DBLP:journals/tifs/FillerJF11}, it is
39 in $O(2^h.n)$ where $h$ is the size of the duplicated
40 matrix. Its complexity is thus negligible compared with the embedding map
48 Thanks to these complexity results, we claim that STABYLO is lightweight.