
Analysis of the accuracy of estimated parameters is an important research direction. In the article, the maximum likelihood estimation is used to estimate CMOS image noise parameters and Fisher information is used to analyse their accuracy. The accuracies of the two parameters are different in different situations. Two applications of it are proposed in this paper. The first one is a guide to image representation. The standard pixel image has higher accuracy for signal-dependent noise and higher error for additive noise, in contrast to the normalised pixel image. Therefore, the corresponding image representation is chosen to estimate the noise parameters according to the dominant noise. The second application of the conclusions is a guide to algorithm design. For standard pixel images, the error of additive noise estimation will largely affect the final denoising result if two kinds of noise are removed simultaneously. Therefore, a divide-and-conquer hybrid total least squares algorithm is proposed for CMOS image restoration. After estimating the parameters, the total least square algorithm is first used to remove the signal-dependent noise of the image. Then, the additive noise parameters of the processed image are updated by using the principal component analysis algorithm, and the additive noise in the image is removed by BM3D. Experiments show that this hybrid method can effectively avoid the problems caused by the inconsistent precision of the two kinds of noise parameters. Compared with the state-of-art methods, the new method shows certain advantages in subjective visual quality and objective image restoration indicators.
Citation: Mingying Pan, Xiangchu Feng. Application of Fisher information to CMOS noise estimation[J]. AIMS Mathematics, 2023, 8(6): 14522-14540. doi: 10.3934/math.2023742
[1] | Saravanan Shanmugam, Mohamed Rhaima, Hamza Ghoudi . Exponential synchronization analysis for complex dynamical networks with hybrid delays and uncertainties under given control parameters. AIMS Mathematics, 2023, 8(12): 28976-29007. doi: 10.3934/math.20231484 |
[2] | Arthit Hongsri, Wajaree Weera, Thongchai Botmart, Prem Junsawang . Novel non-fragile extended dissipative synchronization of T-S fuzzy complex dynamical networks with interval hybrid coupling delays. AIMS Mathematics, 2023, 8(12): 28601-28627. doi: 10.3934/math.20231464 |
[3] | Changping Dai, Weiyuan Ma, Ling Guo . Synchronization of generalized fractional complex networks with partial subchannel losses. AIMS Mathematics, 2024, 9(3): 7063-7083. doi: 10.3934/math.2024344 |
[4] | Pratap Anbalagan, Evren Hincal, Raja Ramachandran, Dumitru Baleanu, Jinde Cao, Chuangxia Huang, Michal Niezabitowski . Delay-coupled fractional order complex Cohen-Grossberg neural networks under parameter uncertainty: Synchronization stability criteria. AIMS Mathematics, 2021, 6(3): 2844-2873. doi: 10.3934/math.2021172 |
[5] | N. Jayanthi, R. Santhakumari, Grienggrai Rajchakit, Nattakan Boonsatit, Anuwat Jirawattanapanit . Cluster synchronization of coupled complex-valued neural networks with leakage and time-varying delays in finite-time. AIMS Mathematics, 2023, 8(1): 2018-2043. doi: 10.3934/math.2023104 |
[6] | Zhifeng Lu, Fei Wang, Yujuan Tian, Yaping Li . Lag synchronization of complex-valued interval neural networks via distributed delayed impulsive control. AIMS Mathematics, 2023, 8(3): 5502-5521. doi: 10.3934/math.2023277 |
[7] | Shuang Li, Xiao-mei Wang, Hong-ying Qin, Shou-ming Zhong . Synchronization criteria for neutral-type quaternion-valued neural networks with mixed delays. AIMS Mathematics, 2021, 6(8): 8044-8063. doi: 10.3934/math.2021467 |
[8] | Li Liu, Yinfang Song, Hong Yu, Gang Zhang . Almost sure exponential synchronization of multilayer complex networks with Markovian switching via aperiodically intermittent discrete observa- tion noise. AIMS Mathematics, 2024, 9(10): 28828-28849. doi: 10.3934/math.20241399 |
[9] | Jingyang Ran, Tiecheng Zhang . Fixed-time synchronization control of fuzzy inertial neural networks with mismatched parameters and structures. AIMS Mathematics, 2024, 9(11): 31721-31739. doi: 10.3934/math.20241525 |
[10] | Rongjie Yu, Hengguo Yu, Min Zhao . Steady states and spatiotemporal dynamics of a diffusive predator-prey system with predator harvesting. AIMS Mathematics, 2024, 9(9): 24058-24088. doi: 10.3934/math.20241170 |
Analysis of the accuracy of estimated parameters is an important research direction. In the article, the maximum likelihood estimation is used to estimate CMOS image noise parameters and Fisher information is used to analyse their accuracy. The accuracies of the two parameters are different in different situations. Two applications of it are proposed in this paper. The first one is a guide to image representation. The standard pixel image has higher accuracy for signal-dependent noise and higher error for additive noise, in contrast to the normalised pixel image. Therefore, the corresponding image representation is chosen to estimate the noise parameters according to the dominant noise. The second application of the conclusions is a guide to algorithm design. For standard pixel images, the error of additive noise estimation will largely affect the final denoising result if two kinds of noise are removed simultaneously. Therefore, a divide-and-conquer hybrid total least squares algorithm is proposed for CMOS image restoration. After estimating the parameters, the total least square algorithm is first used to remove the signal-dependent noise of the image. Then, the additive noise parameters of the processed image are updated by using the principal component analysis algorithm, and the additive noise in the image is removed by BM3D. Experiments show that this hybrid method can effectively avoid the problems caused by the inconsistent precision of the two kinds of noise parameters. Compared with the state-of-art methods, the new method shows certain advantages in subjective visual quality and objective image restoration indicators.
It is now widely admitted by the scientific community that the increase of the human activity over the last century exerts a high pressure on the equilibrium of ecological systems, which can account for a major defaunation and a fast loss of biodiversity, often qualified as the "sixth extinction" [7,12]. One of the main causes of this biological crisis is the degradation of the ecological environments of wildlife, which takes various forms, in the forefront of which are deforestation and habitat fragmentation [8,10,13]. Facing the challenge of restoring biodiversity, while maintaining human activity at a reasonable level, a carefully considered solution consists in the implementation of ecological corridors between each component of the fragmented environment, so as to increase the connectivity of natural habitats and to avoid local extinction of several wildlife species. For example, the restoration of aquatic ecosystems by maintaining hydrological connectivity with riparian corridors is studied in [15]; different complexity and connectivity patterns are investigated at multiple scales in [21].
In this paper, our aim is precisely to propose an original mathematical model, in order to study these ecological concerns. Therefore, we consider a network of ecological systems, which is intended to model the complex dynamics of trophic chains in a degraded area. The complex network structure reproduces the heterogeneous natural environment, which is perturbed by fragmentation, by coupling several patches on which interacting wild species are living. On each patch, the ecological inter-species dynamics are modeled by a Lotka-Volterra predator-prey model with Holling type II functional response, which is able to describe several biological dynamics, such as extinction, coexistence or ecological cycles [9,17,19] (see also e.g. [3,18,23,24] for other Lotka-Volterra type models). Next, the complex network is constructed so that each patch can admit its own dynamic. In this way, the local components of the network can for instance exhibit an extinction equilibrium on some places, whereas other places can present cycles. Furthermore, migrations of biological individuals in space, between each component of the fragmented environment, are taken into account by coupling the patches of the network, as schematized in Figure 1, where the disks model the patches of the fragmented habitat, and the arrows model the ecological corridors between these patches.
With this mathematical model in hand, we analyze the effect of the couplings on the local dynamics of each patch. In particular, we investigate the possibility to modify a local dynamic of extinction of the species, by increasing the couplings with patches on which persistence, with or without oscillations, is ensured. More generally, we search for sufficient conditions of synchronization of the local dynamics, under a variation of the couplings, which roughly means that a unique local dynamic is imposed on each patch. On this point, we establish a novel theorem of near-synchronization, which guarantees that the complex network remains in a neighborhood of a synchronization state, provided the coupling strength is strong enough, even if the local behaviors are non-identical.
Our choice to model the dynamics of interacting species living in a fragmented environment by a complex network structure follows a line of recent works. Indeed, synchronization in complex networks has been studied in a great number of papers, for various real-world applications, among them coupled oscillators, networks of chemical reactions, neural networks or meta-populations models (see for instance [1,2,5,14] and the references therein). Recently, complex networks of Lotka-Volterra models have been studied in [6]; however, the sufficient conditions of synchronization which have been established in this paper, correspond only to the particular case of identical dynamics. This state of the art highlights the main contributions of the present paper: it is the first time, at our knowledge, that a near-synchronization result for complex networks of non-identical systems is established at a theoretical level.
Our paper is organized as follows. In the next Section, we show how to construct a complex network of predator-prey systems, stemming from a Lotka-Volterra model embedded with a Holling type II functional response. In Section 3, we establish our main theoretical result, with Theorem 2, which establishes sufficient conditions of near-synchronization in a network of non-identical systems. Finally, in Section 4, we illustrate our theoretical statements by relevant numerical simulations.
In this section, we present the construction of a complex network of Lotka-Volterra systems, which describes the dynamics of interacting species living in a fragmented environment.
Let us consider a biological environment in which two species interact. We assume that the densities of the species are determined by a predator-prey model of Lotka-Volterra type, which can be written by:
{˙x=rx(1−x)−cxyα+x,˙y=−dy+cxyα+x. | (2.1) |
Here, x and y denote the prey and predator density, respectively; ˙x and ˙y denote their derivatives with respect to the time variable t. The parameters r, c, d and α are positive coefficients; r is the birth rate of the preys, d is the mortality rate of predators, and c, α determine the non-linear interaction between preys and predators. As mentioned in our introduction, the dynamics of the predator-prey system (2.1) have been widely studied (see for instance [11]). Depending on the values of the parameters r, c, d, α, the solutions of system (2.1) can be attracted to a coexistence equilibrium, to an extinction equilibrium or to a limit cycle. The extinction equilibrium is denoted E0=(0,0). Under the parameter conditions c>d and α<c−dd, system (2.1) admits a coexistence equilibrium E1, which implies persistence of each species, given by
E1=(αdc−d,rαc−d(1−αdc−d)). | (2.2) |
System (2.1) also admits the equilibrium E2=(1,0). Let us introduce the critical value α0 given by
α0=c−dc+d. | (2.3) |
It is well-known (see for instance [11], Chapter 3 or [4], Section 3.4.1) that system (2.1) undergoes a Hopf bifurcation at α=α0. For α<α0, a stable limit cycle bifurcates from the persistence equilibrium E1. This Hopf bifurcation is illustrated in Figure 2. Therefore, for α small enough, system (2.1) presents oscillations, which are interpreted as healthy ecological cycles.
Next, we assume that the geographical habitat of the species is perturbed by the anthropic extension, so that it is fragmented in several patches. This fragmentation is likely to alter the equilibrium of the ecological system. In order to model such a fragmented environment, we construct a complex network of predator-prey models as follows.
First, let n>0 denote the number of patches on the fragmented environment. On each patch i∈{1,…,n}, we denote by (xi,yi) the densities of preys and predators respectively. We assume that each patch i∈{1,…,n} can be connected to other patches and we denote by Ni⊂{1,…,n} the set of patches which are connected to patch i. We assume that migrations of biological individuals can occur between two connected patches, at rates σ1 for preys and σ2 for predators. In this way, the dynamics of the fragmented environment are determined by the following complex network:
{˙xi=rixi(1−xi)−cixiyiαi+xi−σ1∑j∈Ni(xi−xj),˙yi=−diyi+cixiyiαi+xi−σ2∑j∈Ni(yi−yj), | (2.4) |
for 1≤i≤n, with σ1≥0 and σ2≥0. Note that the index i ranges over the set {1,…,n}, whereas the subscript j, in the two sums which determine the couplings of the network, ranges of the set Ni of the neighbors of i.
We emphasize that the parameters ri, ci, di, αi can differ from one patch to another, which means that the ecological dynamics are non-identical within the fragmented environment. For instance, some patches could present limit cycles, whereas other patches could exhibit an extinction of both species. Note also that the couplings are symmetric, which means that if the species xi, yi of patch i can move towards some patch j, then the species xj, yj of patch j can conversely move towards patch i.
One remarkable case of fragmented environment is that of a complete graph topology, for which we have Ni={1,…,n}∖{i}; this situation means that each patch is connected to all other patches. At the opposite, if the coupling parameters σ1, σ2 are equal to 0, then no migration of individuals occur in the network.
Let us now introduce some notations. Let X=((x1,y1),…,(xn,yn))⊤∈R2n. For each i∈{1,…,n}, we denote
λi=(ri,ci,di,αi)⊤∈R4,f1(xi,yi,λi)=rixi(1−xi)−cixiyiαi+xi,f2(xi,yi,λi)=−diyi+cixiyiαi+xi,g1(xi,X,σ1)=−σ1∑j∈Ni(xi−xj),g2(yi,X,σ2)=−σ2∑j∈Ni(yi−yj). | (2.5) |
We also denote σ=(σ1,σ2)⊤∈R2 and
Λ=(λ1,…,λn)⊤∈R4n,F(X,Λ)=(f1(x1,y1,λ1),f2(x1,y1,λ1),…,f1(xn,yn,λn),f2(xn,yn,λn))⊤∈R2n,G(X,σ)=(g1(x1,X,σ1),g2(y1,X,σ2),…,g1(xn,X,σ1),g2(yn,X,σ2))⊤∈R2n. | (2.6) |
With these notations, the complex network (2.4) can be written under the following short form
˙X=F(X,Λ)+G(X,σ). | (2.7) |
Our first result guarantees that the complex network (2.7) admits global solutions, whose components are non-negative and bounded.
Theorem 1. Let X0∈(R+)2n. Then the complex network problem determined by equation (2.7) and X(0)=X0 admits a unique global solution X(t,X0) defined on [0,+∞), whose components are non-negative.
Furthermore, the flow induced by equation (2.7) admits a positively invariant region Θ which is compact in (R+)2n.
Before giving the proof of Theorem 1, we recall a Gronwall type inequality, whose proof can be found in [22] for instance.
Lemma 1. Let v denote a continuously differentiable function defined on [0,T], with T>0. Assume that v satisfies
˙v(t)+δv(t)≤γ, |
for all t∈[0,T], with δ>0 and γ>0. Then we have
v(t)≤[v(0)−γδ]e−δt+γδ,t∈[0,T]. |
We continue with the proof of Theorem 1.
Proof of Theorem 1. Let X0∈(R+)2n. Standard results of the theory of differential equations (see for instance [16]) guaranty that the Cauchy problem determined by equation (2.7) and the initial condition X(0)=X0 admits a unique local solution, which we denote X(t,X0), defined on [0,T], with T>0.
Let us now prove that the components of the local solution X(t,X0) are non-negative on [0,T]. First, we recall that the initial condition X0 belongs to (R+)2n. Next, we observe that the equations of system (2.4) can be rewritten
{˙xi=xiϕ1(xi,yi)+σ1∑j∈Nixj,˙yi=yiϕ2(x,i,yi)+σ2∑j∈Niyj, |
with 1≤i≤n and
ϕ1(xi,yi)=xi(1−xi)−ciyiαi+xi−σ1|Ni|,ϕ2(xi,yi)=−diyi+ciyiαi+xi−σ2|Ni|, |
where |Ni| denotes the cardinal of the finite set Ni. Hence, by virtue of Proposition A.17 in [20], the components (xi,yi)1≤i≤n satisfy xi(t)≥0, yi(t)≥0, for t∈[0,T].
Finally, let us prove that the flow induced by equation (2.7) admits a positively invariant region Θ, which is compact in (R+)2n; as a consequence, every local solution X(t,X0) will be global in time. We first remark that for each i∈{1,…,n}, positive coefficients ai, bi can be found such that
ris(1−s)≤ai−bis, |
for all s∈R. Afterwards, we introduce
d0=min1≤i≤ndi,a0=n∑i=1ai,b0=min1≤i≤nbi,c0=min(b0,d0), | (2.8) |
and we denote by P the total population of preys and predators in the complex network:
P(t)=n∑i=1[xi(t)+yi(t)],t∈[0,T]. | (2.9) |
Since the couplings are symmetric, the sum over i∈{1,…,n} of all equations of system (2.4) leads to
˙P(t)=n∑i=1rixi(t)[1−xi(t)]−n∑i=1diyi(t). |
Next, we write
−n∑i=1diyi(t)≤−d0n∑i=1yi(t), |
and analogously
n∑i=1rixi(t)[1−xi(t)]≤n∑i=1[ai−bixi(t)]≤a0−b0n∑i=1xi(t). |
We obtain
˙P(t)+c0P(t)≤a0. |
Applying Lemma 1 leads to
P(t)≤[P(0)−a0c0]e−c0t+a0c0, |
which implies that the region Θ defined by
Θ={X=(xi,yi)1≤i≤n∈(R+)2n;n∑i=1(xi+yi)≤a0c0} | (2.10) |
is a positively invariant and compact region. The proof is complete.
Remark 1. It is worth emphasizing that the bound a0c0 of the positively invariant region Θ defined by (2.10) does not depend on the coupling parameters σ1, σ2. This uniform bound will be of great interest in Section 3 for studying the global dynamics of the complex network (2.7).
Now that the solutions of the complex network (2.7) are proved to be well defined and global in time, we give the definition of synchronization.
Definition 1. Let i,j∈{1,…,n} such that i≠j. We say that the patches i and j of the complex network (2.7) synchronize in Θ if, for any initial condition X0∈Θ, the solution of (2.7) starting from X0 satisfies
limt→+∞(|xi(t)−xj(t)|2+|yi(t)−yj(t)|2)=0. |
We say that the complex network (2.7) synchronizes in Θ if every pair (i,j) of patches synchronizes in Θ.
Remark 2. Note that synchronization can occur in the complex network (2.7) without imposing any particular asymptotic dynamics; for example, the complex network could synchronize towards a global dynamic of extinction, towards a global dynamic of coexistence, or towards a global dynamic of limit cycles (it could even happen that a new dynamic emerges from the complex network structure). We will establish in Section 3 sufficient conditions of synchronization and discover that a complex network of non-identical instances of the predator-prey model (2.1) is likely not to fully synchronize.
In this section, we search for sufficient conditions of synchronization in the complex network (2.7). In a great number of papers, sufficient conditions of synchronization are established in the case of complex networks of identical systems. Here, we investigate the case of non-identical systems. We prove a novel theorem which shows that a complex network of non-identical systems can reach a near-synchronization state. For the sake of simplicity, we assume that the graph underlying the complex network (2.7) is complete, that is, each patch is connected to all other patches; equivalently, we have Ni={1,…,n}∖{i} for 1≤i≤n, where Ni denotes the finite set of patches which are connected to patch i. For all i,j∈{1,…,n}, we introduce the energy functions ui,j defined along the trajectories of the complex network by
ui,j(t)=12[|xi(t)−xj(t)|2+|yi(t)−yj(t)|2], | (3.1) |
and for λi=(ri,di,ci,αi),λj=(rj,dj,cj,αj)∈R4, we denote
‖λi−λj‖∞=max{|ri−rj|,|di−dj|,|ci−cj|,|αi−αj|}. |
The next theorem establishes an estimate of the energy functions ui,j defined by (3.1).
Theorem 2. There exist positive constants η, δ such that, for any initial condition X0∈Θ, the solution of the complex network (2.7), starting from X0, satisfies
˙ui,j(t)≤η‖λi−λj‖∞u1/2i,j(t)+[δ−2(n−1)˜σ]ui,j(t),t>0, | (3.2) |
where ˜σ=min{σ1,σ2}.
Furthermore, the constants η and δ do not depend on the coupling parameters σ1, σ2.
We begin with a technical lemma.
Lemma 2. There exist positive constants kr, 1≤r≤6, such that the functions f1 and f2 defined in (2.5) satisfy
|f1(xi,yi,λi)−f1(xj,yj,λj)|≤k1‖λi−λj‖∞+k2|xi−xj|+k3|yi−yj|,|f2(xi,yi,λi)−f2(xj,yj,λj)|≤k4‖λi−λj‖∞+k5|xi−xj|+k6|yi−yj|, | (3.3) |
for all (xi,yi)1≤i≤n, in the invariant region Θ defined by (2.10) and for all i,j∈{1,…,n}.
Furthermore, the constants kr, 1≤r≤6, do not depend on the coupling parameters σ1, σ2.
Proof of Lemma 2. Since (xi,yi)1≤i≤n belongs to the invariant region Θ defined by (2.10), there exists a positive constant K such that
|xi|≤K,|yi|≤K, |
for all i∈{1,…,n}.
In order to establish the estimates (3.3), we first write
|diyi−djyj|≤|diyi−diyj|+|diyj−djyj|, |
which leads to
|diyi−djyj|≤d+|yi−yj|+K|di−dj|, | (3.4) |
where d+=max1≤i≤ndi. Similarly, we have
|rixi(1−xi)−rjxj(1−xj)|≤|rixi(1−xi)−rixj(1−xj)|+|rixj(1−xj)−rjxj(1−xj)|≤ri|xi(1−xi)−xj(1−xj)|+K(1+K)|ri−rj|. |
Now we write
|xi(1−xi)−xj(1−xj)|≤|xi−xj|+|x2i−x2j|≤|xi−xj|+|xi+xj|×|xi−xj|≤(1+2K)|xi−xj|, |
from which it follows that
|rixi(1−xi)−rjxj(1−xj)|≤r+(1+2K)|xi−xj|+K(1+K)|ri−rj|, | (3.5) |
where r+=max1≤i≤nri. Finally, we compute
|cixiyiαi+xi−cjxjyjαj+xj|≤|(αj+xj)cixiyi−(αi+xi)cjxjyj)(αi+xi)(αj+xj)|≤1αiαj[|αjcixiyi−αicjxjyj|+|cixixjyi−cjxixjyj|]. |
For the first term in the brackets, we write
|αjcixiyi−αicjxjyj|≤|αjcixiyi−αjcixjyj|+|αjcixjyj−αicjxjyj|≤αjci|xiyi−xjyj|+K2|αjci−αicj|≤αjci|xiyi−xiyj|+αjci|xiyj−xjyj|+K2|αjci−αicj|≤αjciK(|yi−yj|+|xi−xj|)+K2αj|ci−cj|+K2cj|αi−αj|. |
For the second term in the brackets, we have
|cixixjyi−cjxixjyj|≤|cixixjyi−cixixjyj|+|cixixjyj−cjxixjyj|≤ciK2|yi−yj|+K3|ci−cj|, |
which leads to
|cixiyiαi+xi−cjxjyjαj+xj|≤α+c+K(α−)2|xi−xj|+a+c+K+c+K2(α−)2|yi−yj|+K2α++K3(α−)2|ci−cj|+c+K2(α−)2|αi−αj|, | (3.6) |
where α+=max1≤i≤nαi, c+=max1≤i≤nci and α−=min1≤i≤nαi.
Gathering inequalities (3.5) and (3.6) leads to
|f1(xi,yi,λi)−f1(xj,yj,λj)|≤k1‖λi−λj‖∞+k2|xi−xj|+k3|yi−yj|, |
whereas gathering inequalities (3.4) and (3.6) leads to
|f2(xi,yi,λi)−f2(xj,yj,λj)|≤k4‖λi−λj‖∞+k5|xi−xj|+k6|yi−yj|, |
with positive constants kr, 1≤r≤6, which do not depend on σ neither on the number n of patches in the network. The proof of Lemma 2 is complete.
We continue with the proof of Theorem 2.
Proof of Theorem 2. We compute the derivative of ui,j(t) along a trajectory (xi(t),yi(t))1≤i≤n of the complex network (2.7), starting from an initial condition (xi,0,yi,0)1≤i≤n∈Θ. In order to lighten our notations, we omit the time dependence:
˙ui,j=(xi−xj)(˙xi−˙xj)+(yi−yj)(˙yi−˙yj)=(xi−xj)[f1(xi,yi,λi)−σ1∑k≠i(xi−xk)−f1(xj,yj,λj)+σ1∑k≠j(xj−xk)]+(yi−yj)[f2(xi,yi,λi)−σ2∑k≠i(yi−yk)−f2(xj,yj,λj)+σ2∑k≠j(yj−yk)]. |
Now, we observe that
σ1∑k≠i(xi−xk)−σ1∑k≠j(xj−xk)=σ1(n−1)(xi−xj),σ2∑k≠i(yi−yk)−σ2∑k≠j(yj−yk)=σ2(n−1)(yi−yj). |
By virtue of Lemma 2, we obtain
˙ui,j≤|xi−xj||k1‖λi−λj‖∞+k2|xi−xj|+k3|yi−yj||−σ1(n−1)(xi−xj)2+|yi−yj||k4‖λi−λj‖∞+k5|xi−xj|+k6|yi−yj||−σ2(n−1)(yi−yj)2≤max{k1,k4}‖λi−λj‖∞×[|xi−xj|+|yi−yj|]+[k2−σ1(n−1)]|xi−xj|2+[k6−σ2(n−1)]|yi−yj|2+(k3+k5)|xi−xj|×|yi−yj|. |
Next, we use the standard inequality a+b≤√2×√a2+b2, which is valid for all a,b∈R+, to write
|xi−xj|+|yi−yj|≤√2√|xi−xj|2+|yi−yj|2≤2√ui,j, |
thus we have
max{k1,k4}‖λi−λj‖∞×[|xi−xj|+|yi−yj|]≤η‖λi−λj‖∞u1/2i,j, |
with η=2×max{k1,k4}. In parallel, the Young inequality a×b≤a22+b22 yields
|xi−xj|×|yi−yj|≤|xi−xj|22+|yi−yj|22≤ui,j. |
Finally, we set δ=2max{k2,k6}+k3+k5 and obtain
˙ui,j≤η‖λi−λj‖∞u1/2i,j+[δ−2(n−1)˜σ]ui,j, |
which completes the proof of Theorem 2.
Let us now discuss on some consequences of Theorem 2. We observe that estimate (3.2) is a differential inequality, which implies that for all i,j∈{1,…,n}, the energy function ui,j is bounded by the solution w of the Bernoulli equation
˙w=η‖λi−λj‖∞w1/2+[δ−2(n−1)˜σ]w. | (3.7) |
If λi=λj, the latter differential equation can be simplified, so that estimate (3.2) becomes
˙ui,j(t)≤[δ−2(n−1)˜σ]ui,j(t), |
which directly yields the following corollary.
Corollary 1. Assume that λi=λj for some i,j∈{1,…,n}. Then the patches i and j synchronize if the following condition is fulfilled:
(n−1)˜σ>δ. | (3.8) |
If λi=λj for all i,j∈{1,…,n}, then obviously the whole network synchronizes under condition (3.8). Next, since the constant δ does not depend on the coupling parameters σ1, σ2, the sufficient condition (3.8) can easily be satisfied, provided the number n of patches in the network is sufficiently large, or provided the minimum coupling strength ˜σ=min{σ1,σ2} is sufficiently large.
From the ecological point of view, increasing the number n of patches in the network would correspond to a worse fragmentation of the habitat, which is not a reasonable strategy for our purposes. However, increasing the minimum coupling strength ˜σ can be realized by providing wider ecological corridors.
The non trivial case of Theorem 2 corresponds to a complex network of non-identical patches, for which we have λi≠λj for at least one pair (i,j)∈{1,…,n}2. In that case, the synchronization state {(xi,yi)=(xj,yj)} is likely to present a soft loss of stability. Indeed, it is well-known that the solution w of the Bernoulli equation (3.7) converges towards a positive limit given by
limt→+∞w(t)=(η‖λi−λj‖∞δ−(n−1)˜σ)2, |
provided w(0)>0. We obtain the following corollary.
Corollary 2. The energy function ui,j defined by (3.1), along with the solution of the complex network (2.7) starting, from X0∈Θ, satisfies
0≤lim supui,j(t)≤(η‖λi−λj‖∞δ−(n−1)˜σ)2. | (3.9) |
Finally, since the constants η and δ do not depend on the coupling parameters σ1, σ2, then a sufficiently large value of the minimum coupling strength ˜σ ensures that any neighborhood of the synchronization state {(xi,yi)=(xj,yj)} can be reached, which justifies the expression near-synchronization. This remark can be formalized by the following definition.
Definition 2. Let i,j∈{1,…,n} such that i≠j. We say that the patches i and j of the complex network (2.7) nearly synchronize in Θ with respect to ˜σ if, for any initial condition X0∈Θ, and for any ε>0, the solution of (2.7) starting from X0 satisfies
0≤limt→+∞(|xi(t)−xj(t)|2+|yi(t)−yj(t)|2)<ε, |
for ˜σ sufficiently large.
We say that the complex network (2.7) nearly synchronizes in Θ if every pair (i,j) of patches nearly synchronizes in Θ.
The discussion above can now be formulated as follows.
Corollary 3. The complex network (2.7) nearly synchronizes in Θ with respect to the minimum coupling strength ˜σ.
The latter corollary means that a sufficiently large minimal coupling strength ˜σ ensures that the dynamics of each patch of the complex network (2.4) will be almost identical. Furthermore, by virtue of (3.9), if the minimal coupling strength ˜σ increases, or if the number of patches n increases, then the near-synchronization state is getting closer to a synchronization state. In other words, ˜σ and n can be viewed as parameters to control the level of near-synchronization in the network.
In this section, we provide two examples that illustrate the near-synchronization pattern of a complex network, under the effect of the couplings. We consider networks with two and ten patches, with different coupling strengths, so as to highlight various emergent dynamics. For the sake of simplicity, we assume that σ1=σ2 and write σ instead of σ1, σ2. Our complete computation codes are provided in the Appendix.
We start by showing the effect of the coupling in a simple two-patches network. Although it may appear simple, the case of a two-patches network is fundamental, since it models a natural habitat which is divided by a single obstacle. Notably, the construction of a single road crossing a forest ecosystem exerts a high perturbation on the biodiversity hosted by this environment. Similarly, the establishment of a single river barrier or of a single maritime dyke profoundly modifies the behavior of a water hosted ecosystem.
Therefore, we consider the system given by
{˙x1=r1x1(1−x1)−c1x1y1α1+x1−σ(x1−x2),˙y1=−d1y1+c1x1y1α1+x1−σ(y1−y2),˙x2=r2x2(1−x2)−c2x2y2α2+x2−σ(x2−x1),˙y2=−d2y2+c2x2y2α2+x2−σ(y2−y1), | (4.1) |
with the parameter values from Table 1 and σ>0. On patch 1, the parameters are chosen so that the system converges to a coexistence steady state (α=0.5) or to a limit cycle (α=0.05): the birth rate r1 of preys and the death rate d1 of predators are of the same order; in both cases, extinction is avoided. On patch 2, the parameters are chosen so that the system converges to the extinction state (it suffices to decrease the birth rate of preys: r1=0.2).
Patch 1 | Patch 2 | |||
Parameter | Value | Parameter | Value | |
r1 | 1 | r2 | 0.2 | |
d1 | 1 | d2 | 1 | |
c1 | 2 | c2 | 3 | |
α1 | 0.5 or 0.05 | α2 | 0.05 |
The initial conditions are defined by
(x1(0),y1(0))=(0.3,0.2),(x2(0),y2(0))=(0.5,0.3). |
We show in Figure 3 the corresponding orbits of the two-patches complex network (4.1), in absence of coupling (σ=0), and in Figure 4 the orbits for a positive coupling strength (σ=0.3); two cases are investigated: in the first case, we have α1=0.5, whereas in the second case, we have α1=0.05. Let us discuss the numerical results in regard of Theorem 2.
For σ=0 and α1=0.5 (Figure 3(a)), the orbit (x1,y1) of the first patch is attracted to a persistence equilibrium, while the orbit (x2,y2) of the second patch is attracted to the extinction equilibrium. For σ=0 and α1=0.05 (Figure 3(b)), the orbit of the second patch is the same, but the orbit (x1,y1) of the first patch is now attracted to a limit cycle. We can predict, by virtue of Theorem 2, that the two-patches network will present a near-synchronization dynamic if the coupling strength σ is sufficiently large. However, we aim to describe the common dynamic which is reached under this near-synchronization pattern. Indeed, for σ=0.3, Figure 4 shows that the near-synchronization can hide various emergent dynamics that might occur. For σ=0.3 and α1=0.5 (Figure 4(a)), the two patches nearly synchronize towards a persistence equilibrium, whereas for σ=0.3 and α1=0.05 (Figure 4(b)), the two patches nearly synchronize towards a limit cycle. In both cases, the dynamics are obviously modified: the level of persistence is decreased on patch 1 for α1=0.5 (Figure 4(a)), as well as the amplitude of the oscillations on patch 1 for α1=0.05 (Figure 4(b)); however, the extinction on patch 2 is avoided. Overall, these results that show the good health of the ecosystem (persistence or ecological oscillations) can be recovered by the setting of a strong connection between the patches of the fragmented habitat.
Finally, we aim to experiment an increase of the number n of patches in the network. Therefore, we consider a ten-patches network with a complete graph topology; we choose randomly a set of parameters for each patch, such that
0≤ri≤1,0≤di≤1,0≤ci≤3,0.01≤αi≤0.41, |
for each i∈{1,2,…,10}; the initial conditions are also chosen randomly in the interval [0.1,0.6]. In this way, the patches of the complex network exhibit various local dynamics in absence of coupling. The corresponding orbits are show in Figure 5; the corresponding time series are also provided in Figure 6. For instance, the orbit (x3,y3) of patch 3 is attracted to a limit cycle; the orbit (x4,y4) of patch 4 converges to the equilibrium E2(1,0) (persistence of the preys and extinction of the predators); the orbit (x6,y6) of patch 6 converges to the extinction equilibrium.
We depict in Figure 7 the orbits of the same ten-patches complex network, with a relatively large coupling strength (σ=1); the corresponding time series are again provided in Figure 8. According to Theorem 2, we observe that the orbits nearly synchronize. The coupling strength is large enough to ensure that the near-synchronization is very closed to a synchronization state. Furthermore, the near-synchronization is characterized by a limit cycle; the amplitude of that limit cycle can slightly vary from one patch to another (for instance, y7 is less than 0.7, while y8 reaches 0.8).
From the ecological point of view, these results show again that healthy biological oscillations can be recovered by the setting of numerous and efficient connections between the patches of a fragmented environment.
In this paper, we proposed a complex network to model a heterogeneous geographical habitat of species which is perturbed by an anthropic extension, being fragmented in several patches, where the fragmentation is likely to alter the equilibrium of the ecological system. The complex network was constructed by coupling several patches on which interacting wild species are living and where, for each patch, the ecological inter-species dynamics were modeled by a Lotka-Volterra predator-prey model with Holling type II functional response. An important feature of the complex network is that each patch can admit its own dynamic and migrations of biological individuals in space, between each component of the fragmented environment, are taken into account by coupling the patches of the network. We proved new sufficient conditions for the near-synchronization of the complex network, which guarantees that the complex network remains in a neighborhood of a synchronization state, provided the coupling strength is strong enough, even if the local behaviors are non-identical. This result allows us to modify the local dynamic of extinction of the species, by increasing the couplings with patches on which persistence, with or without oscillations, is ensured.
When proving the sufficient conditions of synchronization, we discovered that a complex network of non-identical instances of the predator-prey model (2.1) is likely not to fully synchronize. This feature motivates the setting of an optimal control problem, so as to exert a command on the dynamics of the complex network (2.7) and to reach a synchronization state, even in the case of non-identical patches. As future work, we intend to apply optimal control theory to remedy the default of synchronization, where the coupling strengths will be represented by external controls acting on the dynamics of the network.
Another interesting research perspective corresponds to the fact that the dispersion of the species from one patch to another could be modeled at a refined level by adding time delays in the coupling terms; in particular, it will be natural to investigate the impact of time delays on the near synchronization dynamic that we have established in the present paper.
This work was partially supported by Portuguese funds through CIDMA, The Center for Research and Development in Mathematics and Applications of University of Aveiro, and the Portuguese Foundation for Science and Technology (FCT–Fundação para a Ciência e a Tecnologia), within project UIDB/04106/2020. Silva was also supported by the FCT Researcher Program CEEC Individual 2018 with reference CEECIND/00564/2018.
All authors declare no conflicts of interest in this paper.
In this appendix, we provide the computation codes of our numerical simulations. These codes are written in Python3 and require the scientific libraries matplotlib, numpy and scipy.
Computation code for a two-patches network
#!/usr/bin/env python3 # Scientific libraries from matplotlib import pyplot as plt import numpy as np from scipy.integrate import odeint from random import random # Parameters r1 = 1; d1 = 1; c1 = 2; alpha1 = 0.05 # or 0.5 sigma = 0.3 # or 0 r2 = 0.2; d2 = 1; c2 = 3; alpha2 = 0.05 def lotka(X, t): x1, y1, x2, y2 = X dx1 = r1*x1*(1-x1) - c1*x1*y1/(alpha1+x1) - sigma*(x1-x2) dy1 = -d1*y1 + c1*x1*y1/(alpha1+x1) - sigma*(y1-y2) dx2 = r2*x2*(1-x2) - c2*x2*y2/(alpha2+x2) + sigma*(x1-x2) dy2 = -d2*y2 + c2*x2*y2/(alpha2+x2) + sigma*(y1-y2) return [dx1, dy1, dx2, dy2] # Phase portrait time = np.arange(0, 50, 0.01) plt.figure() x10 = 0.3; y10 = 0.2; x20 = 0.5; y20 = 0.3 orbit = odeint(lotka, [x10, y10, x20, y20], time) x1, y1, x2, y2 = orbit.T plt.plot(x1, y1, 'b', lw = 0.5) plt.plot(x2, y2, 'r', lw = 1) plt.show() |
Computation code for a ten-patches network
#!/usr/bin/env python3 # Scientific libraries from matplotlib import pyplot as plt import numpy as np from scipy.integrate import odeint from random import random # Parameters r1 = random(); d1 = random(); c1 = 3*random(); alpha1 = 0.01 + 0.4*random() r2 = random(); d2 = random(); c2 = 3*random(); alpha2 = 0.01 + 0.4*random() r3 = random(); d3 = random(); c3 = 3*random(); alpha3 = 0.01 + 0.4*random() r4 = random(); d4 = random(); c4 = 3*random(); alpha4 = 0.01 + 0.4*random() r5 = random(); d5 = random(); c5 = 3*random(); alpha5 = 0.01 + 0.4*random() r6 = random(); d6 = random(); c6 = 3*random(); alpha6 = 0.01 + 0.4*random() r7 = random(); d7 = random(); c7 = 3*random(); alpha7 = 0.01 + 0.4*random() r8 = random(); d8 = random(); c8 = 3*random(); alpha8 = 0.01 + 0.4*random() r9 = random(); d9 = random(); c9 = 3*random(); alpha9 = 0.01 + 0.4*random() r10 = random(); d10 = random(); c10 = 3*random(); alpha10 = 0.01 + 0.4*random() def lotka(X, t): x1, y1, x2, y2, x3, y3, x4, y4, x5, y5, x6, y6, x7, y7, x8, y8, x9, y9, x10, y10 = X dx1 = r1*x1*(1-x1) - c1*x1*y1/(alpha1+x1) - sigma*(9*x1-x2-x3-x4-x5-x6-x7-x8-x9-x10) dy1 = -d1*y1 + c1*x1*y1/(alpha1+x1) - sigma*(9*y1-y2-y3-y4-y5-y6-y7-y8-y9-y10) dx2 = r2*x2*(1-x2) - c2*x2*y2/(alpha2+x2) - sigma*(9*x2-x1-x3-x4-x5-x6-x7-x8-x9-x10) dy2 = -d2*y2 + c2*x2*y2/(alpha2+x2) - sigma*(9*y2-y1-y3-y4-y5-y6-y7-y8-y9-y10) dx3 = r3*x3*(1-x3) - c3*x3*y3/(alpha3+x3) - sigma*(9*x3-x1-x2-x4-x5-x6-x7-x8-x9-x10) dy3 = -d3*y3 + c3*x3*y3/(alpha3+x3) - sigma*(9*y3-y2-y1-y4-y5-y6-y7-y8-y9-y10) dx4 = r4*x4*(1-x4) - c4*x4*y4/(alpha4+x4) - sigma*(9*x4-x1-x2-x3-x5-x6-x7-x8-x9-x10) dy4 = -d4*y4 + c4*x4*y4/(alpha4+x4) - sigma*(9*y4-y2-y3-y1-y5-y6-y7-y8-y9-y10) dx5 = r5*x5*(1-x5) - c5*x5*y5/(alpha5+x5) - sigma*(9*x5-x2-x3-x4-x1-x6-x7-x8-x9-x10) dy5 = -d5*y5 + c5*x5*y5/(alpha5+x5) - sigma*(9*y5-y2-y3-y4-y1-y6-y7-y8-y9-y10) dx6 = r6*x6*(1-x6) - c6*x6*y6/(alpha6+x6) - sigma*(9*x6-x2-x3-x4-x5-x1-x7-x8-x9-x10) dy6 = -d6*y6 + c6*x6*y6/(alpha6+x6) - sigma*(9*y6-y2-y3-y4-y5-y1-y7-y8-y9-y10) dx7 = r7*x7*(1-x7) - c7*x7*y7/(alpha7+x7) - sigma*(9*x7-x2-x3-x4-x5-x6-x1-x8-x9-x10) dy7 = -d7*y7 + c7*x7*y7/(alpha7+x7) - sigma*(9*y7-y2-y3-y4-y5-y6-y1-y8-y9-y10) dx8 = r8*x8*(1-x8) - c8*x8*y8/(alpha8+x8) - sigma*(9*x8-x2-x3-x4-x5-x6-x7-x1-x9-x10) dy8 = -d8*y8 + c8*x8*y8/(alpha8+x8) - sigma*(9*y8-y2-y3-y4-y5-y6-y7-y1-y9-y10) dx9 = r9*x9*(1-x9) - c9*x9*y9/(alpha9+x9) - sigma*(9*x9-x2-x3-x4-x5-x6-x7-x8-x1-x10) dy9 = -d9*y9 + c9*x9*y9/(alpha9+x9) - sigma*(9*y9-y2-y3-y4-y5-y6-y7-y8-y1-y10) dx10 = r10*x10*(1-x10) - c10*x10*y10/(alpha10+x10) - sigma*(9*x10-x2-x3-x4-x5-x6-x7-x8-x9-x1) dy10 = -d10*y10 + c10*x10*y10/(alpha10+x10) - sigma*(9*y10-y2-y3-y4-y5-y6-y7-y8-y9-y1) return [dx1, dy1, dx2, dy2, dx3, dy3, dx4, dy4, dx5, dy5, dx6, dy6, dx7, dy7, dx8, dy8, dx9, dy9, dx10, dy10] # Phase portrait time = np.arange(0,200, 0.01) plt.figure() X0 = [0.1 + 0.5*random() for k in range(20)] sigma = 0 # or 1 orbit = odeint(lotka, X0, time) x1, y1, x2, y2, x3, y3, x4, y4, x5, y5, x6, y6, x7, y7, x8, y8, x9, y9, x10, y10 = orbit.T plt.plot(x1, y1) plt.plot(x2, y2) plt.plot(x3, y3) plt.plot(x4, y4) plt.plot(x5, y5) plt.plot(x6, y6) plt.plot(x7, y7) plt.plot(x8, y8) plt.plot(x9, y9) plt.plot(x10, y10) plt.show() |
[1] |
C. Tomasi, R. Manduchi, Bilateral filtering for gray and color images, Sixth international conference on computer vision (IEEE Cat. No. 98CH36271), (1998), 839–846. http://dx.doi.org/10.1109/ICCV.1998.710815 doi: 10.1109/ICCV.1998.710815
![]() |
[2] |
S. Osher, M. Burger, D. Goldfarb, J. Xu, E. Yin, An iterative regularization method for total variation-based image restoration, Multiscale Model. Sim., 4 (2005), 460–489. http://dx.doi.org/10.1137/040605412 doi: 10.1137/040605412
![]() |
[3] |
Y. zhang, S. Li, B. Wu, S. Du, Image multiplicative denoising using adaptive Euler's elastica as the regularization, J. Sci. Comput., 90 (2022), 69. http://dx.doi.org/10.1007/s10915-021-01721-7 doi: 10.1007/s10915-021-01721-7
![]() |
[4] | L. Rudin, P. L. Lions, S. Osher, Geometric level set methods in imaging, vision, and graphics, New York, USA: Springer, 2003. http://dx.doi.org/10.1007/0-387-21810-6_6 |
[5] |
J. Shi, S. Osher, A nonlinear inverse scale space method for a convex multiplicative noise model, SIAM J. Imaging Sci., 1 (2008), 294–321. http://dx.doi.org/10.1137/070689954 doi: 10.1137/070689954
![]() |
[6] |
K. Bredies, K. Kunisch, T. Pock, Total generalized variation, SIAM J. Imaging Sci., 3 (2010), 492–526. http://dx.doi.org/10.1137/090769521 doi: 10.1137/090769521
![]() |
[7] |
Y. Lv, Total generalized variation denoising of speckled images using a primal-dual algorithm, J. Appl. Math. Comput., 62 (2020), 489–509. http://dx.doi.org/10.1007/s12190-019-01293-8 doi: 10.1007/s12190-019-01293-8
![]() |
[8] |
A. Ben-Loghfyry, A. Hakim, A. Laghrib, A denoising model based on the fractional Beltrami regularization and its numerical solution, J. Appl. Math. Comput., 69 (2023), 1431–1463. http://dx.doi.org/10.1007/s12190-022-01798-9 doi: 10.1007/s12190-022-01798-9
![]() |
[9] |
T. H. Ma, T. Z. Huang, X. L. Zhao, Spatially dependent regularization parameter selection for total generalized variation-based image denoising, Comput. Appl. Math., 37 (2018), 277–296. http://dx.doi.org/10.1007/s40314-016-0342-8 doi: 10.1007/s40314-016-0342-8
![]() |
[10] |
H. Houichet, A. Theljani, M. Moakher, A nonlinear fourth-order PDE for image denoising in Sobolev spaces with variable exponents and its numerical algorithm, Comput. Appl. Math., 40 (2021), 1–29. http://dx.doi.org/10.1007/s40314-021-01462-1 doi: 10.1007/s40314-021-01462-1
![]() |
[11] |
A. Hakim, A. Ben-Loghfyry, A total variable-order variation model for image denoising, AIMS Math., 4 (2019), 1320–1335. http://dx.doi.org/10.3934/math.2019.5.1320 doi: 10.3934/math.2019.5.1320
![]() |
[12] |
J. L. Starck, E. J. Candès, D. L. Donoho, The curvelet transform for image denoising, IEEE T. Image Process., 11 (2002), 670–684. http://dx.doi.org/10.1109/TIP.2002.1014998 doi: 10.1109/TIP.2002.1014998
![]() |
[13] |
J. Yang, Y. Wang, W. Xu, Q. Dai, Image and video denoising using adaptive dual-tree discrete wavelet packets, IEEE T. Circ. Syst. Vid., 19 (2009), 642–655. http://dx.doi.org/10.1109/TCSVT.2009.2017402 doi: 10.1109/TCSVT.2009.2017402
![]() |
[14] |
L. Fan, X. Li, H. Fan, Y. Feng, C. Zhang, Adaptive texture-preserving denoising method using gradient histogram and nonlocal self-similarity priors, IEEE T. Circ. Syst. Vid., 29 (2019), 3222–3235. http://dx.doi.org/10.1109/TCSVT.2018.2878794 doi: 10.1109/TCSVT.2018.2878794
![]() |
[15] | Z. Long, N. H. Younan, Denoising of images with multiplicative noise corruption, 2005 13th European Signal Processing Conference, (2005), 1–4. |
[16] |
W. Dong, L. Zhang, G. Shi, X. Li, Nonlocally centralized sparse representation for image restoration, IEEE T. Image Process., 22 (2012), 1620–1630. http://dx.doi.org/10.1109/TIP.2012.2235847 doi: 10.1109/TIP.2012.2235847
![]() |
[17] |
A. Buades, B. Coll, J. M. Morel, A non-local algorithm for image denoising, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05), (2005), 60–65. http://dx.doi.org/10.1109/CVPR.2005.38 doi: 10.1109/CVPR.2005.38
![]() |
[18] |
K. Dabov, A. Foi, V. Katkovnik, K. Egiazarian, Image denoising by sparse 3-D transform-domain collaborative filtering, IEEE T. Image Process., 16 (2007), 2080–2095. http://dx.doi.org/10.1109/TIP.2007.901238 doi: 10.1109/TIP.2007.901238
![]() |
[19] |
S. Gu, L. Zhang, W. Zuo, X. Feng, Weighted nuclear norm minimization with application to image denoising, 2014 IEEE Conference on Computer Vision and Pattern Recognition, (2014), 2862–2869. http://dx.doi.org/10.1109/CVPR.2014.366 doi: 10.1109/CVPR.2014.366
![]() |
[20] |
J. Mairal, F. Bach, J. Ponce, G. Sapiro, A. Zisserman, Non-local sparse models for image restoration, 2009 IEEE 12th International Conference on Computer Vision, (2009), 2272–2279. http://dx.doi.org/10.1109/ICCV.2009.5459452 doi: 10.1109/ICCV.2009.5459452
![]() |
[21] |
W. Dong, L. Zhang, G. Shi, X. Li, Nonlocally centralized sparse representation for image restoration, IEEE T. Image Process., 22 (2013), 1620–1630. http://dx.doi.org/10.1109/TIP.2012.2235847 doi: 10.1109/TIP.2012.2235847
![]() |
[22] |
Q. Guo, C. Zhang, Y. Zhang, H. Liu, An efficient SVD-based method for image denoising, IEEE T. Circ. Syst. Vid., 26 (2016), 868–880. http://dx.doi.org/10.1109/TCSVT.2015.2416631 doi: 10.1109/TCSVT.2015.2416631
![]() |
[23] |
M. Yahia, T. Ali, M. M. Mortula, R. Abdelfattah, S. E. Mahdy, N. S. Arampola, Enhancement of SAR speckle denoising using the improved iterative filter, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 13 (2020), 859–871. http://dx.doi.org/10.1109/JSTARS.2020.2973920 doi: 10.1109/JSTARS.2020.2973920
![]() |
[24] |
K. Zhang, W. Zuo, Y. Chen, D. Meng, L. Zhang, Beyond a gaussian denoiser: Residual learning of deep CNN for image denoising, IEEE T. Image Process., 26 (2017), 3142–3155. http://dx.doi.org/10.1109/TIP.2017.2662206 doi: 10.1109/TIP.2017.2662206
![]() |
[25] |
Y. Meng, J. Zhang, A novel gray image denoising method using convolutional neural network, IEEE Access, 10 (2022), 49657–49676. http://dx.doi.org/10.1109/ACCESS.2022.3169131 doi: 10.1109/ACCESS.2022.3169131
![]() |
[26] |
G. Wang, Z. Pan, Z. Zhang, Deep CNN Denoiser prior for multiplicative noise removal, Multimed. Tools Appl., 78 (2019), 29007–29019. http://dx.doi.org/10.1007/s11042-018-6294-9 doi: 10.1007/s11042-018-6294-9
![]() |
[27] |
H. Tian, B. Fowler, A. E. Gamal, Analysis of temporal noise in CMOS photodiode active pixel sensor, IEEE J. Solid-St. Circ., 36 (2001), 92–101. http://dx.doi.org/10.1109/4.896233 doi: 10.1109/4.896233
![]() |
[28] |
J. Zhang, K. Hirakawa, Improved denoising via Poisson mixture modeling of image sensor noise, IEEE T. Image Process., 26 (2017), 1565–1578. http://dx.doi.org/10.1109/TIP.2017.2651365 doi: 10.1109/TIP.2017.2651365
![]() |
[29] |
D. Chen, X. Teng, Novel variational approach for generalized signal dependent noise removal, 2018 11th International Symposium on Computational Intelligence and Design (ISCID), 2 (2018), 380–384. http://dx.doi.org/10.1109/ISCID.2018.10187 doi: 10.1109/ISCID.2018.10187
![]() |
[30] | J. Zhang, Y. Duan, Y. Lu, M. K. Ng, H. Chang, Bilinear constraint based ADMM for mixed Poisson-Gaussian noise removal, arXiv preprint arXiv: 1910.08206, 2019. |
[31] |
M. Ghulyani, M. Arigovindan, Fast roughness minimizing image restoration under mixed Poisson–Gaussian noise, IEEE T. Image Process., 30 (2021), 134–149. http://dx.doi.org/10.1109/TIP.2020.3032036 doi: 10.1109/TIP.2020.3032036
![]() |
[32] |
S. Huang, T. Lu, Z. Lu, J. Rong, X. Zhao, J. Li, CMOS image sensor fixed pattern noise calibration scheme based on digital filtering method, Microelectron. J., 124 (2022), 10543. http://dx.doi.org/10.1016/j.mejo.2022.105431 doi: 10.1016/j.mejo.2022.105431
![]() |
[33] |
S. Lee, M. G. Kang, Poisson-Gaussian noise reduction for X-Ray images based on local linear minimum mean square error shrinkage in nonsubsampled contourlet transform domain, IEEE Access, 9 (2021), 100637–100651. http://dx.doi.org/10.1109/ACCESS.2021.3097078 doi: 10.1109/ACCESS.2021.3097078
![]() |
[34] |
J. Zhang, K. Hirakawa, Improved denoising via Poisson mixture modeling of image sensor noise, IEEE T. Image Process., 26 (2017), 1565–1578. http://dx.doi.org/10.1109/TIP.2017.2651365 doi: 10.1109/TIP.2017.2651365
![]() |
[35] | A. Repetti, E. Chouzenoux, J. Pesquet, A penalized weighted least squares approach for restoring data corrupted with signal-dependent noise, 2012 Proceedings of the 20th European Signal Processing Conference (EUSIPCO), (2012), 1553–1557. |
[36] |
K. Hirakawa, T. W. Parks, Image denoising using total least squares, IEEE T. Image Process., 15 (2006), 2730–2742. http://dx.doi.org/10.1109/TIP.2006.877352 doi: 10.1109/TIP.2006.877352
![]() |
[37] |
Y. Qiu, Z. Gan, Y. Fan, X. Zhu, An adaptive image denoising method for mixture Gaussian noise, 2011 International Conference on Wireless Communications and Signal Processing (WCSP), (2011), 1–5. http://dx.doi.org/10.1109/WCSP.2011.6096774 doi: 10.1109/WCSP.2011.6096774
![]() |
[38] |
J. Byun, S. Cha, T. Moon, FBI-Denoiser: Fast blind image denoiser for Poisson-Gaussian noise, IEEE Conference on Computer Vision and Pattern Recognition, (2021), 5768–5777. http://dx.doi.org/10.1109/CVPR46437.2021.00571 doi: 10.1109/CVPR46437.2021.00571
![]() |
[39] |
D. L. Donoho, De-noising by soft-thresholding, IEEE T. Inform. Theory, 41 (1995), 613–627. https://doi.org/10.1109/18.382009 doi: 10.1109/18.382009
![]() |
[40] |
J. Immerkær, Fast noise variance estimation, Comput. Vis. Image Und., 64 (1996), 300–302. http://dx.doi.org/10.1006/cviu.1996.0060 doi: 10.1006/cviu.1996.0060
![]() |
[41] |
D. Zoran, Y. Weiss, Scale invariance and noise in natural images, 2009 IEEE 12th International Conference on Computer Vision, (2009), 2209–2216. https://doi.org/10.1109/ICCV.2009.5459476 doi: 10.1109/ICCV.2009.5459476
![]() |
[42] |
S. Zhu, Z. Yu, Self-guided filter for image denoising, IET Image Prosess, 14 (2020), 2561–2566. http://dx.doi.org/10.1049/iet-ipr.2019.1471 doi: 10.1049/iet-ipr.2019.1471
![]() |
[43] |
L. Lu, W. Jin, X. Wang, Non-local means image denoising with a soft threshold, IEEE Signal Proc. Lett., 22 (2015), 833–837. http://dx.doi.org/10.1109/LSP.2014.2371332 doi: 10.1109/LSP.2014.2371332
![]() |
[44] |
D. -G. Kim, Y. Ali, M. A. Farooq, A. Mushtaq, M. A. A. Rehman, Z. H. Shamsi, Hybrid Deep Learning Framework for Reduction of Mixed Noise via Low Rank Noise Estimation, IEEE Access, 10 (2022), 46738–46752. http://dx.doi.org/10.1109/ACCESS.2022.3170490 doi: 10.1109/ACCESS.2022.3170490
![]() |
[45] |
D. H. Shin, R. H. Park, S. Yang, J. H. Jung, Block-based noise estimation using adaptive Gaussian filtering, IEEE T. Consum. Electr., 51 (2005), 218–226. http://dx.doi.org/10.1109/TCE.2005.1405723 doi: 10.1109/TCE.2005.1405723
![]() |
[46] |
A. Danielyan, A. Foi, Noise variance estimation in nonlocal transform domain, 2009 International Workshop on Local and Non-Local Approximation in Image Processing, (2009), 41–45. https://doi.org/10.1109/LNLA.2009.5278404 doi: 10.1109/LNLA.2009.5278404
![]() |
[47] |
X. Liu, M. Tanaka, M. Okutomi, Noise level estimation using weak textured patches of a single noisy image, 2012 19th IEEE International Conference on Image Processing, (2012), 665–668. http://dx.doi.org/10.1109/ICIP.2012.6466947 doi: 10.1109/ICIP.2012.6466947
![]() |
[48] |
X. Liu, M. Tanaka, M. Okutomi, Estimation of signal dependent noise parameters from a single image, 2013 IEEE International Conference on Image Processing, (2013), 79–82. http://dx.doi.org/10.1109/ICIP.2013.6738017 doi: 10.1109/ICIP.2013.6738017
![]() |
[49] |
C. Sutour, A.-C. Deledalle, F.-J. Aujol, Estimation of the noise level function based on a non-parametric detection of homogeneous image regions, SIAM J. Imaging Sci., 8 (2015), 1–31. http://dx.doi.org/10.1137/15M1012682 doi: 10.1137/15M1012682
![]() |
[50] |
Z. Wang, Z. Huang, Y. Xu, Y. Zhang, X. Li, X. Li, et al., Image Noise Level Estimation by Employing Chi-Square Distribution, 2021 IEEE 21st International Conference on Communication Technology (ICCT), (2021), 1158–1161. http://dx.doi.org/10.1109/ICCT52962.2021.9657946 doi: 10.1109/ICCT52962.2021.9657946
![]() |
[51] |
V. A. Pimpalkhute, R. Page, A. Kothari, K. M. Bhurchandi, V. M. Kamble, Digital image noise estimation using DWT coefficients, IEEE T. Image Process., 30 (2021), 1962–1972. http://dx.doi.org/10.1109/TIP.2021.3049961 doi: 10.1109/TIP.2021.3049961
![]() |
[52] |
J. Sijbers, A. den Dekker, Maximum likelihood estimation of signal amplitude and noise variance from MR data, Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine, 51 (2004), 586–594. http://dx.doi.org/10.1002/mrm.10728 doi: 10.1002/mrm.10728
![]() |
[53] |
M. W. Wu, Y. Jin, Y. Li, T. Song, P. Y. Kam, Maximum-likelihood, magnitude-based, amplitude and noise variance estimation, IEEE Signal Proc. Lett., 28 (2021), 414–418. http://dx.doi.org/10.1109/LSP.2021.3055464 doi: 10.1109/LSP.2021.3055464
![]() |
[54] |
A. G. Vostretsov, S. G. Filatova, The Estimation of Parameters of Pulse Signals Having an Unknown Form That Are Observed against the Background of the Additive Mixture of the White Gaussian Noise and a Linear Component with Unknown Parameters, Journal of Communications Technology and Electronics, 66 (2021), 938–947. http://dx.doi.org/10.1134/S106422692108009X doi: 10.1134/S106422692108009X
![]() |
[55] |
R. A. Fisher, On the mathematical foundations of theoretical statistics, Philosophical Transactions of the Royal Society of London. Series A, Containing Papers of a Mathematical or Physical Character, 222 (1922), 309–368. http://dx.doi.org/10.1098/rsta.1922.0009 doi: 10.1098/rsta.1922.0009
![]() |
[56] |
Y. Ouyang, S. Wang, L. Zhang, Quantum optical interferometry via the photon-added two-mode squeezed vacuum states, J. Opt. Soc. Am. B, 33 (2016), 1373–1381. https://doi.org/10.1364/JOSAB.33.001373 doi: 10.1364/JOSAB.33.001373
![]() |
[57] |
G. C. Knee, W. J. Munro, Fisher information versus signal-to-noise ratio for a split detector, PHYS. REV. A, 92 (2015), 012130. http://dx.doi.org/10.1103/PhysRevA.92.012130 doi: 10.1103/PhysRevA.92.012130
![]() |
[58] |
J. Chao, E. S. Ward, R. J. Ober, Fisher information theory for parameter estimation in single molecule microscopy: tutorial, J. Opt. Soc. Am. A, 33 (2016), B36–B57. https://doi.org/10.1364/JOSAA.33.000B36 doi: 10.1364/JOSAA.33.000B36
![]() |
[59] |
J. Kirkpatrick, R. Pascanu, N. Rabinowitz, J. Veness, G. Desjardins, A. A. Rusu, et al., Overcoming catastrophic forgetting in neural networks, Proceedings of the national academy of sciences, 114 (2017), 3521–3526. https://doi.org/10.1073/pnas.1611835114 doi: 10.1073/pnas.1611835114
![]() |
[60] | J. Martens, New insights and perspectives on the natural gradient method, The Journal of Machine Learning Research, 21 (2020), 5776–5851. |
1. | M A Aziz-Alaoui, Guillaume Cantin, Alexandre Thorel, Synchronization of Turing patterns in complex networks of reaction–diffusion systems set in distinct domains, 2024, 37, 0951-7715, 025011, 10.1088/1361-6544/ad1a48 |
Patch 1 | Patch 2 | |||
Parameter | Value | Parameter | Value | |
r1 | 1 | r2 | 0.2 | |
d1 | 1 | d2 | 1 | |
c1 | 2 | c2 | 3 | |
α1 | 0.5 or 0.05 | α2 | 0.05 |