Loading [MathJax]/extensions/TeX/boldsymbol.js
Research article

Four mathematical modeling forms for correlation filter object tracking algorithms and the fast calculation for the filter

  • The correlation filter object tracking algorithm has gained extensive attention from scholars in the field of tracking because of its excellent tracking performance and efficiency. However, the mathematical modeling relationships of correlation filter tracking frameworks are unclear. Therefore, many forms of correlation filters are susceptible to confusion and misuse. To solve these problems, we attempted to review various forms of the correlation filter and discussed their intrinsic connections. First, we reviewed the basic definitions of the circulant matrix, convolution, and correlation operations. Then, the relationship among the three operations was discussed. Considering this, four mathematical modeling forms of correlation filter object tracking from the literature were listed, and the equivalence of the four modeling forms was theoretically proven. Then, the fast solution of the correlation filter was discussed from the perspective of the diagonalization property of the circulant matrix and the convolution theorem. In addition, we delved into the difference between the one-dimensional and two-dimensional correlation filter responses as well as the reasons for their generation. Numerical experiments were conducted to verify the proposed perspectives. The results showed that the filters calculated based on the diagonalization property and the convolution property of the cyclic matrix were completely equivalent. The experimental code of this paper is available at https://github.com/110500617/Correlation-filter/tree/main.

    Citation: Yingpin Chen, Kaiwei Chen. Four mathematical modeling forms for correlation filter object tracking algorithms and the fast calculation for the filter[J]. Electronic Research Archive, 2024, 32(7): 4684-4714. doi: 10.3934/era.2024213

    Related Papers:

    [1] T. A. Shaposhnikova, M. N. Zubova . Homogenization problem for a parabolic variational inequality with constraints on subsets situated on the boundary of the domain. Networks and Heterogeneous Media, 2008, 3(3): 675-689. doi: 10.3934/nhm.2008.3.675
    [2] Ken-Ichi Nakamura, Toshiko Ogiwara . Periodically growing solutions in a class of strongly monotone semiflows. Networks and Heterogeneous Media, 2012, 7(4): 881-891. doi: 10.3934/nhm.2012.7.881
    [3] Junlong Chen, Yanbin Tang . Homogenization of nonlinear nonlocal diffusion equation with periodic and stationary structure. Networks and Heterogeneous Media, 2023, 18(3): 1118-1177. doi: 10.3934/nhm.2023049
    [4] Benjamin Contri . Fisher-KPP equations and applications to a model in medical sciences. Networks and Heterogeneous Media, 2018, 13(1): 119-153. doi: 10.3934/nhm.2018006
    [5] Thomas Geert de Jong, Georg Prokert, Alef Edou Sterk . Reaction–diffusion transport into core-shell geometry: Well-posedness and stability of stationary solutions. Networks and Heterogeneous Media, 2025, 20(1): 1-14. doi: 10.3934/nhm.2025001
    [6] Iryna Pankratova, Andrey Piatnitski . Homogenization of convection-diffusion equation in infinite cylinder. Networks and Heterogeneous Media, 2011, 6(1): 111-126. doi: 10.3934/nhm.2011.6.111
    [7] Feiyang Peng, Yanbin Tang . Inverse problem of determining diffusion matrix between different structures for time fractional diffusion equation. Networks and Heterogeneous Media, 2024, 19(1): 291-304. doi: 10.3934/nhm.2024013
    [8] Xavier Blanc, Claude Le Bris . Improving on computation of homogenized coefficients in the periodic and quasi-periodic settings. Networks and Heterogeneous Media, 2010, 5(1): 1-29. doi: 10.3934/nhm.2010.5.1
    [9] Avner Friedman . PDE problems arising in mathematical biology. Networks and Heterogeneous Media, 2012, 7(4): 691-703. doi: 10.3934/nhm.2012.7.691
    [10] Bendong Lou . Self-similar solutions in a sector for a quasilinear parabolic equation. Networks and Heterogeneous Media, 2012, 7(4): 857-879. doi: 10.3934/nhm.2012.7.857
  • The correlation filter object tracking algorithm has gained extensive attention from scholars in the field of tracking because of its excellent tracking performance and efficiency. However, the mathematical modeling relationships of correlation filter tracking frameworks are unclear. Therefore, many forms of correlation filters are susceptible to confusion and misuse. To solve these problems, we attempted to review various forms of the correlation filter and discussed their intrinsic connections. First, we reviewed the basic definitions of the circulant matrix, convolution, and correlation operations. Then, the relationship among the three operations was discussed. Considering this, four mathematical modeling forms of correlation filter object tracking from the literature were listed, and the equivalence of the four modeling forms was theoretically proven. Then, the fast solution of the correlation filter was discussed from the perspective of the diagonalization property of the circulant matrix and the convolution theorem. In addition, we delved into the difference between the one-dimensional and two-dimensional correlation filter responses as well as the reasons for their generation. Numerical experiments were conducted to verify the proposed perspectives. The results showed that the filters calculated based on the diagonalization property and the convolution property of the cyclic matrix were completely equivalent. The experimental code of this paper is available at https://github.com/110500617/Correlation-filter/tree/main.



    In this paper, we study the positive solutions of the periodic-parabolic problem

    {ut=μk(x,t)Δu+m(x,t)uc(x,t)up, in Ω×R,uν=0, on Ω×R,u(x,0)=u(x,T), in Ω, (1.1)

    where Ω is a bounded domain of RN(N1) with smooth boundary Ω, ν is the outward normal vector of Ω, μ>0 and p>1 is constant, m(x,t)Cα,α2(ˉΩ×R)(0<α<1) is T-periodic in t, k(x,t), c(x,t)Cα,1(ˉΩ×R) are positive and T-periodic in t. It is known that the periodic reaction-diffusion equation (1.1) can be accurately used to describe different diffusion phenomena in infectious diseases, microbial growth, and population ecology, see [1,2,3,4]. From a biological point of view, Ω represents the habitat of species u and μk(x,t) stands for the diffusion rate, which is time and space dependent. The function m(x,t) represents the growth rate of species. In this situation, in the subset {(x,t)Ω×R:m(x,t)>0}, the species will increase, while in {(x,t)Ω×R:m(x,t)<0}, species will decrease. The coefficient c(x,t) means that environment Ω can accommodate species u. There are many interesting conclusions about the study of the reaction-diffusion equation, see [5,6,7,8] for the elliptic problems and [9,10,11,12,13,14] for the periodic problems.

    In particular, if k(x,t)k(t) for xˉΩ, problem (1.1) has been well investigated by Hess [2], Cantrell and Cosner [1]. Let λ(μ) be the unique principal eigenvalue of the eigenvalue problem

    {utμk(t)Δum(x,t)u=λ(μ)u, in Ω×R,uν=0, on Ω×R,u(x,0)=u(x,T), in Ω.

    It follows from [2,15] that Eq (1.1) has a positive periodic solution θμ(x,t) if and only if λ(μ)<0. In addition, Dancer and Hess [16] and Daners and López-Gómez [17] studied the effect of μ on the positive periodic solution of Eq (1.1) with various boundary conditions. The most interesting conclusion of [11,16,17] is that

    limμ0+θμ(x,t)=θ(x,t) locally uniformly  in Ω×[0,T],

    here θ(x,t) is the maximum nonnegative periodic solution of

    {ut=m(x,t)uc(x,t)up,tR,u(x,0)=u(x,T).

    However, there is little result on the associated large diffusion and the effect of large diffusion on positive solutions.

    Our goal is to study the existence and uniqueness of positive periodic solutions of Eq (1.1) and the asymptotic behavior of positive periodic solutions when the diffusion rate μ is large. To this end, let λ(μ;m) be the principal eigenvalue of

    {utμk(x,t)Δum(x,t)u=λ(μ;m)u, in Ω×R,uν=0, on Ω×R,u(x,0)=u(x,T), in Ω. (1.2)

    {It is well known that λ(μ;m) plays a major role in the study of the positive periodic solution of Eq (1.1). The properties of λ(μ;m) will be established in Section 2. In addition, let W2,pν(Ω)={uW2,p(Ω):uν=0}(N<p<). If u0W2,pν(Ω), it follows from [2] that the semilinear initial value problem}

    {ut=μk(x,t)Δu+m(x,t)uc(x,t)up, in Ω×R,uν=0, on Ω×R,u(x,0)=u0(x), in Ω,

    has a unique solution U(x,t)=U(x,t;u0) satisfying

    U(x,t)C1+α,1+α2(ˉΩ×[0,T])C2+α,1+α2(ˉΩ×(0,T]).

    Our first result is the existence and uniqueness of positive periodic solutions of Eq (1.1). For simplicity, in the rest of this paper, we use the following notations:

    k(t)=Ω1k(x,t)dx,m(t)=Ωm(x,t)k(x,t)dx,c(t)=Ωc(x,t)k(x,t)dx.

    Theorem 1.1. Suppose that T0m(t)dt>0. Then Eq (1.1) admits a unique positive periodic solution θμ(x,t) for all μ>0.

    Remark 1.1. By the result of Section 3, we know that there exists a unique positive solution to Eq (1.1) if and only if λ(μ;m)<0. In the case T0m(t)dt>0, we obtain that λ(μ;m)<0.

    Next, we study the asymptotic behavior of positive periodic solutions when the diffusion rate is large.

    Theorem 1.2. Suppose that

    m(t)>0 for t[0,T]. (1.3)

    Let θμ(x,t) be the unique positive periodic solution of Eq (1.1) for μ>0. Then we have

    limμθμ(x,t)=ω(t) in C1,12(ˉΩ×[0,T]), (1.4)

    where ω(t) is the unique positive periodic solution of

    {ut=m(t)k(t)uc(t)k(t)up,tR,u(0)=u(T). (1.5)

    Remark 1.2. With the approach of local upper-lower solutions developed by Daners and López-Gómez [17] in the study of classical periodic-parabolic logistic equations, we can prove that

    limμ0+θμ(x,t)=θ(x,t) locally uniformly  in Ω×[0,T],

    provided maxΩT0m(x,t)dt>0. It also shows that when m(t)<0<T0maxΩm(x,t)dt, populations with small dispersal rates survive, while populations with large dispersal rates perish. This means that a small diffusion rate is a better strategy than a large diffusion rate under appropriate circumstances.

    The rest of this paper is arranged as follows: In Section 2, we study the properties of principal eigenvalues for the periodic eigenvalue problems. In Section 3, we mainly study the existence, uniqueness and stability of the positive solution to Eq (1.1). Moreover, we investigate the asymptotic profiles of the positive periodic solution to Eq (1.1) as μ in Section 4.

    In this section, we consider the principal eigenvalue of Eq (1.2). To this end, we first study the linear initial value problem

    {utμk(x,t)Δua(x,t)u=0, in Ω×(τ,T],uν=0, on Ω×(τ,T],u(x,τ)=u0(x), in Ω, (2.1)

    where 0τ<T, u0W2,pν(Ω)(N<p<) and a(x,t)Cα,α2(ˉΩ×[τ,)). It is well known that there is a one-to-one correspondence between Eq (2.1) and the evolution operator Uμ(t,τ). Then we can define that u(x,t)=Uμ(t,τ)u0 is the solution of Eq (2.1). For simplicity, let X=Lp(Ω)(N<p<), X1=W2,pν(Ω) and

    F={uCα,α2(ˉΩ×R):u(,t+T)=u(,t) in R}.

    Inspired by the classical works of Hess [2], we first give some important results of Eq (2.1), which will be used in the rest of this paper.

    Lemma 2.1. If u0X is positive, then Uμ(t,τ)u0>0 in C1ν(ˉΩ) for 0τ<tT.

    Proof. Note that X1 is compactly embedded in X. The operator Uμ(t,τ)/X1:X1X1 can be continuously extended to the positive operator Uμ(t,τ)L(X,X1). Thus Uμ(t,τ)u00 in X1. Since sUμ(s,τ)u0 is continuous from [τ,T] to X1 and Uμ(τ,τ)u0=u00, we can get that Uμ(s,τ)u0>0 in X1 as s>τ goes to τ. In addition, we have

    Uμ(t,τ)u0=Uμ(t,s)Uμ(s,τ)u0,

    for τ<s<t. Thus it can be obtained that Uμ(t,τ)u0>0 for 0τ<tT.

    We now study the periodic-parabolic eigenvalue problem

    {utμk(x,t)Δua(x,t)u=λ(μ;a)u, in Ω×(0,T],uν=0, on Ω×(0,T],u(x,0)=u(x,T), in Ω. (2.2)

    If there is a nontrivial solution u(x,t) of Eq (2.2), then λ(μ;a) is called the eigenvalue. In particular, if u(x,t) is positive, then λ(μ;a) is the principal eigenvalue.

    Theorem 2.1. Let Kμ:=Uμ(T,0) and r be the spectral radius of Kμ. Then r is the principal eigenvalue of Kμ with positive eigenfunction u0 if and only if λ(μ;a)=1Tlnr is the principal eigenvalue of Eq (2.2) with positive eigenfunction u(x,t)=eλ(μ;a)tUμ(t,0)u0.

    Proof. It can be proved by the similar arguments as in [2, Proposition 14.4]. For the completeness, we provide a proof in the following. Suppose that r is the principal eigenvalue of Kμ with positive eigenfunction u0X1. Let u(x,t)=eλ(μ;a)tUμ(t,0)u0. Then u(x,t) satisfies

    {utμk(x,t)Δua(x,t)u=λ(μ;a)u, in Ω×(0,T],uν=0, on Ω×(0,T],u(x,0)=u0=1rKμu0=eλ(μ;a)TKμu0=u(x,T), in Ω.

    According to the regularity results, we have u(x,t)C2+α,1+α2(ˉΩ×R). This means that μ=1Tlnr is the principal eigenvalue of Eq (2.2), while u(x,t)=eλ(μ;a)tUμ(t,0)u0 is the corresponding positive eigenfunction.

    On the contrary, suppose that λ(μ;a)=1Tlnr is the eigenvalue of Eq (2.2) with positive eigenfunction u(x,t). Set v(x,t)=eλ(μ;a)tu(x,t). Then v(x,t) is the solution of

    {vtμk(x,t)Δva(x,t)v=0, in Ω×(0,T],vν=0, on Ω×(0,T],v(x,0)=u(x)=:u0, in Ω.

    Thus, we obtain v(x,t)=Uμ(t,0)u0 for 0tT and u0X1 is positive. Hence,

    v(T)=eλ(μ;a)Tu0=Kμu0.

    It follows from Krein-Rutman theorem that eλ(μ;a)T=r.

    Remark 2.1. For τ<t, it follows that Uμ(t,τ) is a compact and strongly positive operator on X1. Moreover, by Krein-Rutman theorem, we obtain r>0, and r is the unique principal eigenvalue of Kμ. This implies that Eq (2.2) has the unique principal eigenvalue λ(μ;a) for any μ>0.

    Lemma 2.2. Let a1(x,t), a(x,t)F satisfy

    a1(x,t)<a2(x,t) in ˉΩ×[0,T].

    Then λ(μ;a2)<λ(μ;a1) for any μ>0.

    Proof. Assume that there exists μ1>0 such that λ(μ1;a2)λ(μ1;a1). Let u1(x,t) and u2(x,t) be corresponding positive eigenfunctions, chosen in such a way that

    0<u1(x,t)<u2(x,t) in ˉΩ×[0,T].

    Then ω(x,t)=u2(x,t)u1(x,t) satisfies

    {ωtμ1k(x,t)Δωa1(x,t)ω>λ(μ1;a1)ω, in Ω×(0,T],ων=0, on Ω×(0,T],ω(x,0)=ω(x,T), in Ω.

    Set ϕ(x,t)=eλ(μ1;a1)tω(x,t), then we have

    {ϕtμ1k(x,t)Δϕa1(x,t)ϕ>0, in Ω×(0,T],ϕν=0, on Ω×(0,T],ϕ(x,0)=ω(x,0)=ω(x,T), in Ω.

    Thus, for any xΩ, we can obtain

    ϕ(x,T)>Kμ1ω(x,0) and ϕ(x,T)=eλ(μ1;a1)Tω(x,0).

    Hence, we obtain

    (eλ(μ1;a1)TKμ1)ω(x,0)>0 in X1.

    Note that ω(x,0)>0, it follows from [2, Theorem 7.3] that

    eλ(μ1;a1)T=rμ1<eλ(μ1;a1)T,

    where rμ1 is the principal eigenvalue of Kμ1. This is a contradiction.

    Lemma 2.3. Suppose that for any nN, an(x,t)F satisfies

    limnan(x,t)=a(x,t) in C1(ˉΩ×[0,T]).

    Then for fixed μ>0, we have

    limnλ(μ;an)=λ(μ;a).

    Proof. For any given ε>0, there exists nεN such that for any n>nε, there holds

    a(x,t)ε<an(x,t)<a(x,t)+ε in ˉΩ×[0,T].

    Notice that λ(μ;a±ε)=λ(μ;a)ε. From Lemma 2.2, we have

    λ(μ;a)ε<λ(μ;an)<λ(μ;a)+ε,

    for any n>nε.

    Lemma 2.4. Let λ(μ;a) be the principal eigenvalue of Eq (2.2) for μ>0. Then we have

    λ(μ;a)T0a(t)dtT0k(t)dt, (2.3)

    here a(t)=Ωa(x,t)k(x,t)dx.

    Proof. First, we consider the case

    T0Ωkt(x,t)k2(x,t)dxdt0.

    Let φ(x,t) be the positive eigenfunction corresponding to the principal eigenvalue λ(μ;a). Taking α>0 satisfies

    lnα=T0Ωkt(x,t)lnφ(x,t)k2(x,t)dxdtT0Ωkt(x,t)k2(x,t)dxdt.

    Then φα:=αφ(x,t) is also the principal eigenfunction of Eq (2.2). It is easy to obtain

    λ(μ;a)T0Ω1k(x,t)dxdt=T0Ωa(x,t)k(x,t)dxdtμT0ΩΔφαφαdxdt=T0Ωa(x,t)k(x,t)dxdtμT0Ω|Dφα|2φ2αdxdt. (2.4)

    This implies that Eq (3.2) holds.

    Next, we consider the case of

    T0Ωkt(x,t)k2(x,t)dxdt=0.

    We can find smooth T-periodic functions {kn(x,t)} such that

    limnkn(x,t)=k(x,t) in C(ˉΩ×[0,T]),

    and

    T0Ω(kn(x,t))tk2n(x,t)dxdt0.

    It follows from Lemma 2.3 that

    limnλn(μ;a)=λ(μ;a),

    where λn(μ;a) is the principal eigenvalue of Eq (2.2) with k(x,t) replaced by kn(x,t). It is clear from Eq (2.4) that

    λn(μ;a)=T0Ωa(x,t)kn(x,t)dxdtμT0Ω|Dφα|2φ2αdxdt.

    Letting n, we have Eq (3.2).

    Remark 2.2. In Eq (3.2), we obtain upper estimates for the principal eigenvalue of the Neumann problem Eq (2.2). Indeed, let λD be the principal eigenvalue of the eigenvalue problem

    {utμk(x,t)Δua(x,t)u=λDu, in Ω×(0,T],u=0, on Ω×(0,T],u(x,0)=u(x,T), in Ω.

    By a similar way as in [2], we can show

    λD1TT0[μk(x,s)+a(s)]ds,

    for any μ>0.

    In this section, we study the existence and uniqueness of positive solutions of Eq (1.1). First, we show that if Eq (1.2) has negative principal eigenvalues, then Eq (1.1) has a unique positive solution. To this end, we recall the upper-lower solutions of Eq (1.1). For the sake of convenience, let

    QT=Ω×(0,T],Q1=Ω×(0,T].

    Definition 3.1. The continuous function ˉu(x,t) is called the upper-solution of Eq (1.1), if

    {utμk(x,t)Δu+m(x,t)uc(x,t)up,inQT,uν0,onQ1,u(x,0)u(x,T),inΩ,

    is satisfied.

    The definition of the lower-solution is similar. We then can prove the following result, see [2,4,5,15].

    Theorem 3.1. Suppose that ˉu(x,t), u_(x,t) are a pair of ordered bounded upper-lower solutions of Eq (1.1). Then Eq (1.1) has a unique positive periodic solution θμ(x,t)C1+α,(1+α)/2(ˉQT) that satisfies

    u_(x,t)θμ(x,t)ˉu(x,t) in ˉQT.

    Proof. Let

    f(x,t,u)=m(x,t)uc(x,t)up.

    Then there exists a constant K>0 such that

    |f(x,t,u2)f(x,t,u1)|K|u2u1|,

    for any (x,t,ui)ˉQT×[u_(x,t),ˉu(x,t)], i=1,2. It follows from Lp theory that for any uCα,α2(ˉQT) satisfying [u_(x,t),ˉu(x,t)], the linear initial value problem

    {vtμk(x,t)Δv+Kv=Ku+f(x,t,u), in QT,vν=0, on Q1,v(x,0)=u(x,T), in Ω,

    admits a unique solution v. Thus, a nonlinear operator v=Fu is defined. We will prove that there is a fixed point for F in four steps.

    Step1. In this step, we prove that if u_u1u2ˉu, then u_v1=Fu1v2=Fu2ˉu.

    Take ω1=v2v1, then ω1 satisfies

    {[ω1]tμk(x,t)Δω1+Kω1=K(u2u1)+f(x,t,u2)f(x,t,u1)0, in QT,ω1ν=0, on Q1,ω1(x,0)=u2(x,T)u1(x,T)0, in Ω.

    By the comparison principle, we obtain ω10. This implies Fu2Fu1. Similarly, let ω2=v1u_, then ω2 satisfies

    {[ω2]tμk(x,t)Δω2+Kω2=K(u1u_)+f(x,t,u1)f(x,t,u_)0, in QT,ω2ν=0, on Q1,ω1(x,0)=u1(x,T)u_(x,T)0, in Ω.

    Thus, u_v1. Similarly, v2ˉu.

    Step2. In this step, we construct a convergent monotone sequence.

    The iterative sequences {un} and {vn} are constructed as follows:

    u1=Fˉu,u2=Fu1,,un=Fun1,
    v1=Fu_,v2=Fv1,,vn=Fvn1.

    Since u_ˉu and F is monotonically non-decreasing, then

    u_v1u1ˉu.

    Similarly, we obtain

    u_vnunˉu.

    And since u_u1ˉu,

    u_u2u1ˉu.

    By induction, we have un+1un. In the same way, we obtain vnvn+1. Thus, we have

    u_v1v2u2u1ˉu.

    This also implies that {un} and {vn} are monotonically bounded sequences, so there are u0(x,t) and v0(x,t) such that

    limnun(x,t)=u0(x,t),limnvn(x,t)=v0(x,t).

    Thus

    u_(x,t)v0(x,t)u0(x,t)ˉu(x,t) in ˉQT.

    Step3. In this step, we prove that u0(x,t) and v0(x,t) are solutions of Eq (1.1).

    Take E=W2,1p(QT)(p>n+2). First, we prove that F:DC(ˉQT) is a compact operator, where

    D={u(x,t)E:u_(x,t)u(x,t)ˉu(x,t) in ˉQT}.

    For u1, u2E, let v1=Fu1 and v2=Fu2, then ω3=v2v1 satisfies

    {[ω3]tμk(x,t)Δω3+Kω3=K(u2u1)+f(x,t,u2)f(x,t,u1), in QT,ω3ν=0, on Q1,ω1(x,0)=u2(x,T)u1(x,T), in Ω.

    By the Lp estimate and embedding theorem, it follows that

    ω3C1+α,1+α2(ˉQT)Cω3W2,1p(QT)C1(u2u1Lp(QT)+u2(x,T)u1(x,T)Lp(Ω)),

    here C and C1 are positive constants. Thus F:DC(ˉQT) is continuous. It is known from the embedding theorem that if u is bounded in W2,1p(QT), then Fu is bounded in C1+α,(1+α)/2(ˉQT). This means that F:DC(ˉQT) is a compact operator.

    Since un is bounded, un=Fun1 has a convergent subsequence in C(ˉQT). By the monotonicity of un,

    limnun(x,t)=u0(x,t) in C(ˉQT).

    Thus u0(x,t) is the solution of Eq (1.1) in W2,1p(QT). The embedding theorem is used again to get u0(x,t)C1+α,(1+α)/2(ˉQT). In the same way, we get that v0(x,t) is the classical solution of Eq (1.1).

    Step4. In this step, we prove the uniqueness and periodicity of the solution of Eq (1.1).

    Since k(x,t), m(x,t) and c(x,t) are periodic about t, then τ(x,t)=u0(x,t+T)u0(x,t) satisfies

    {τt(x,t)μk(x,t)Δτ(x,t)=m(x,t)τ(x,t)pc(x,t)˜up1(x,t)τ(x,t), in QT,τν=0, on Q1,τ(x,0)=0, in Ω, (3.1)

    here ˜u(x,t) is between u0(x,t+T) and u0(x,t). It is well known that the solution of Eq (3.1) is unique, thus u0(x,t+T)u0(x,t) in ˉQT.

    The uniqueness of the solution is based on the results of [2,4] and can also be found in recent research results [11,15]. Assume that v1 and v2 are two positive periodic solutions of Eq (1.1). We first prove that there exists a large constant M>1 such that

    M1v1<v2<Mv1 in QT.

    Indeed, it is clear that there exists M1>1 such that

    v2(x,0)=v2(x,T)<M1v1(x,T)=M1v1(x,0) in Ω.

    This implies v2(x,0)M1v1(x,0) on ˉΩ. Let η(x,t):=M1v1(x,t)v2(x,t), then η(x,t) satisfies

    ηtμk(x,t)Δη=m(x,t)ηc(x,t)[M1vp1vp2]>m(x,t)ηc(x,t)[(M1v1)pvp2]=m(x,t)ηpc(x,t)ςp1(x,t)η,

    where ς(x,t) is lying between M1v1(x,t) and v2(x,t). Notice that ην=0 on Q1. By the maximum principle, we have η>0 in ˉQT. Similarly, we can obtain that there exists M2>0 such that v1<M2v2 in ˉQT. Take M=max{M1,M2}, then we have

    M1v1<v2<Mv1 in QT.

    We know that Mv1(x,t) and M1v1(x,t) are a pair of upper-lower solutions of Eq (1.1). According to the second step, Eq (1.1) has a minimum solution u and a maximum solution u, which satisfies uvu in ˉQT for all solution v satisfying M1v1vMv1. Thus, we obtain uviu for i=1,2. Hence, uu implies the uniqueness of the solution to Eq (1.1). Set

    α=inf{α>0|u(x,t)αu(x,t) in ˉQT}.

    It is clear that α1. Note that if α=1, then u(x,t)u(x,t) in ˉQT. Assume that α>1. Let π(x,t)=αuu. It is known from the maximum principle that π(x,t)>0 in ˉQT. On the other hand, we know that

    π(x,0)=π(x,T)α1u(x,T)=α1u(x,0) on ˉΩ,

    for some small α1>0. We can use the previous method to prove the existence of M to show that

    π(x,t)α1u(x,t) on ˉQT.

    Then we have u(x,t)(αα)u(x,t). This is in contradiction with the definition of α. Thus, we obtain α=1. The uniqueness is proved.

    Lemma 3.1. If λ(μ;m)<0, then Eq (1.1) admits a unique positive periodic solution θμ(x,t)C1+α,(1+α)/2(ˉQT). Moreover, θμ(x,t) is globally asymptotically stable.

    Proof. Let θ(x,t) be a principal eigenfunction of Eq (1.2) normalized by θ(x,t)C(ˉQT)=1. Then u_=εθ(x,t) is a lower-solution of Eq (1.1) for any

    0<ε[λ(μ;m)maxˉQTc(x,t)]1p.

    Take

    M>max{1,[λ(μ;m)minˉQTc(x,t)]1p}.

    Then we have ˉu=Mθ(x,t) is an upper-solution of Eq (1.1). From Theorem 3.1, we get that Eq (1.1) has a unique positive solution θμ(x,t).

    Since θμ(x,t) is the solution of Eq (1.1), then λ(μ;mcθp1μ)=0. Let λ1 be the eigenvalue of the linear problem

    {utμk(x,t)Δu[m(x,t)pc(x,t)θp1μ]u=λ1u, in Ω×(0,T],uν=0, on Ω×(0,T],u(x,0)=u(x,T), in Ω.

    Due to

    m(x,t)c(x,t)ˆup1>m(x,t)pc(x,t)ˆup1,

    for u>0. Then λ1>0. It follows from Theorem 2.1 that r<1. Thus, θμ(x,t) is locally asymptotically stable. In addition, we can choose ε small enough and M large enough such that εθ(x,t) and Mθ(x,t) are a pair of ordered bounded upper-lower solutions of Eq (1.1). Then we know that θμ(x,t) is globally asymptotically stable by the standard iteration argument as in [2].

    Lemma 3.2. If (1.1) has a positive periodic solution, then λ(μ;m)<0.

    Proof. Let θμ(x,t) be a positive periodic solution of Eq (1.1). Thanks to [2], we can get that Eq (1.1) is equivalent to

    (IKμ)θμ(x,0)=T0Uμ(T,τ)c(x,τ)θpμ(x,τ)dτ in X1.

    Notice that θμ(x,t)>0. We now apply [2, Theorem 7.3] to obtain

    eλ(μ;m)T>1.

    Thus, λ(μ;m)<0.

    Proposition 3.1. If T0m(t)dt>0, then Eq (1.1) admits a unique positive periodic solution θμ(x,t) for all μ>0.

    Proof. According to Lemma 2.4, we know that

    λ(μ;m)T0m(t)dtT0k(t)dt. (3.2)

    Due to T0m(t)dt>0, λ(μ;m)<0. This together with Lemma 3.1 implies that Eq (1.1) admits a unique positive periodic solution for all μ>0.

    In this section, we study the asymptotic behavior of the positive periodic solution of Eq (1.1) when the diffusion rate is large. Here we use regularity estimates together with the perturbation technique to prove our main result. To do this, we first consider the perturbation equation

    {ut=μk(x,t)Δu+m(x,t)(u+ε)c(x,t)up, in QT,uν=0, on Q1,u(x,0)=u(x,T), in Ω, (4.1)

    where the parameter ε>0.

    Lemma 4.1. Assume that Eq (1.3) holds. Then Eq (4.1) has a positive periodic solution θεμ(x,t) for μ>0, provided ε>0 is small. Moreover, we can find μ1>0 such that

    limε0+θεμ(x,t)=θμ(x,t) in C1,12(ˉQT), (4.2)

    for μμ1.

    Proof. Through a similar argument as in Theorem 3.1, we know the existence of the positive solution θεμ(x,t) to Eq (4.1). We only need to prove Eq (4.2). Let σ(x,t)=θεμ(x,tμ). Then σ(x,t) satisfies

    {σt=k(x,tμ)Δσ+1μ[m(x,tμ)(σ+ε)c(x,tμ)σp], in QT,σν=0, on Q1,σ(x,0)=θεμ(x,0), in Ω.

    It is known from the Lp estimate that there exists μ1>0 such that σ(x,t) is bounded in W2,1p(Ω×(0,μT]) for any μ>μ1. It then follows that θεμ(x,t) is bounded in W2,1p(QT) for any μ>μ1. Then, taking p large enough, we know from the embedding theorem that θεμ(x,t) is compact in C1,12(ˉQT). Thus there is a subsequence, still denoted by θεμ(x,t), such that

    limε0+θεμ(x,t)=ν(x,t) in C1,12(ˉQT), (4.3)

    for some nonnegative periodic function ν(x,t)C(ˉQT). It follows from the argument of Lemma 3.1 that εθ(x,t) is a lower-solution of Eq (4.1). Thus we have ν(x,t)>0 for (x,t)ˉΩ×[0,T]. Since θεμ(x,t) is bounded and Eq (4.3), ν satisfies

    ν(x,t)=ν(x,0)+μt0[k(x,s)Δν(x,s)+m(x,s)νc(x,s)νp]ds.

    It is easy to obtain

    {νt=μk(x,t)Δν+m(x,t)νc(x,t)νp, in QT,νν=0, on Q1,ν(x,0)=ν(x,T), in Ω.

    By standard parabolic regularity, we know that ν(x,t)C1+α,(1+α)/2(ˉQT). The uniqueness of the solution means that Eq (4.2) holds.

    At the end of this section, we prove Theorem 1.2.

    Proof of Theorem 1.2. We divide the proof into the following three steps.

    Step1. In this step, we prove that θμ(x,t) has a convergent subsequence as μ.

    It follows from a similar argument to Lemma 4.1 that there exists μ1>0 such that θμ(x,t) is compact in C1,12(ˉQT) for any μ>μ1. Thus, by passing to a subsequence, there is a nonnegative periodic function θC(ˉQT) such that

    limμθμ(x,t)=θ(x,t) in C1,12(ˉQT).

    Step2. In this step, we show that θ(x,t) is independent of t.

    Let f(t) be a smooth T-periodic function, then we have

    T0θμ(x,t)ft(t)dt=μT0k(x,t)f(t)Δθμ(x,t)dt+T0m(x,t)θμ(x,t)f(t)dtT0c(x,t)θpμ(x,t)f(t)dt.

    By dividing μ and making μ, we have

    T0k(x,t)f(t)Δθ(x,t)dt=0.

    Since f(t) is arbitrary, we obtain

    Δθ(x,t)=0.

    Then we derive

    Ω|θ(x,t)|2dx=0.

    Thus we have θ(x,t)θ(t) for xˉΩ.

    Step3. In this step, we show that θ(t)=ω(t) in [0,T].

    First, we assert that θ(t)C1((0,)). Indeed, it is easy to obtain from Eq (1.1) that

    t+εtΩus(x,s)k(x,s)dxds=t+εtΩm(x,s)k(x,s)u(x,s)dxdst+εtΩc(x,s)k(x,s)up(x,s)dxds.

    Then we have

    Ωu(x,t+ε)k(x,t+ε)dxΩu(x,t)k(x,t)dx+t+εtΩkt(x,s)k2(x,s)u(x,s)dxds=t+εtΩm(x,s)k(x,s)u(x,s)dxdst+εtΩc(x,s)k(x,s)up(x,s)dxds.

    Taking μ, we obtain

    Ωθ(t+ε)k(x,t+ε)dxΩθ(t)k(x,t)dx+t+εtΩkt(x,s)k2(x,s)θ(s)dxds=t+εtΩm(x,s)k(x,s)θ(s)dxdst+εtΩc(x,s)k(x,s)θp(s)dxds.

    Thus, we derive

    [θ(t)Ω1k(x,t)dx]t=Ωm(x,t)k(x,t)dxθ(t)Ωc(x,t)k(x,t)dxθp(t)Ωkt(x,t)k2(x,t)dxθ(t).

    Hence,

    θt(t)=1k(t)Ωm(x,t)k(x,t)dxθ(t)1k(t)Ωc(x,t)k(x,t)dxθp(t)1k(t)Ωkt(x,t)k2(x,t)dxθ(t)[k(t)]tk(t)θ(t),

    for t>0. Thus θ(t)C1((0,)) holds.

    We then prove that θ(t) satisfies Eq (1.5). It is obvious from Eq (1.1) that

    Ωut(x,t)k(x,t)dx=Ωm(x,t)k(x,t)u(x,t)dxΩc(x,t)k(x,t)up(x,t)dx. (4.4)

    Similarly, suppose that f(t) is a smooth T-periodic function. Multiplying f(t) on both sides of Eq (4.4) and integrating over [0,T] gives

    T0Ωu(x,t)[f(t)k(x,t)]tdxdt=T0Ωm(x,t)k(x,t)u(x,t)f(t)dxdtT0Ωc(x,t)k(x,t)up(x,t)f(t)dxdt.

    Letting μ, we know

    T0Ωθ(t)[f(t)k(x,t)]tdxdt=T0Ωm(x,t)k(x,t)θ(t)f(t)dxdtT0Ωc(x,t)k(x,t)θp(t)f(t)dxdt.

    This implies

    T0Ωf(t)k(x,t)θt(t)dxdt=T0Ωm(x,t)k(x,t)θ(t)f(t)dxdtT0Ωc(x,t)k(x,t)θp(t)f(t)dxdt.

    By the arbitrary of f(t), it follows that

    Ω1k(x,t)dxθt(t)=Ωm(x,t)k(x,t)dxθ(t)Ωc(x,t)k(x,t)dxθp(t).

    Thus, we have

    {θt=˜M(t)˜k(t)θ˜C(t)˜k(t)θp,tR,θ(0)=θ(T).

    Finally, we prove that θ(t)>0 in t[0,T]. We define θεμ(x,t) as the unique positive periodic solution of Eq (4.1) for small ε>0 and large μ. Similarly to the previous argument, it can be shown that

    limμθεμ(x,t)=ωε(t) in C1,12(ˉQT), (4.5)

    where ωε(t) satisfies

    {ωεt=M(t)k(t)(ωε+ε)c(t)k(t)(ωε)p,tR,ωε(0)=ωε(T). (4.6)

    Since

    Ωm(x,t)k(x,t)dx>0 in [0,T],

    we can obtain that Eq (1.5) admits a unique periodic positive solution ω(t). Note that ω(t) is a lower solution of Eq (4.6). Then there exists a unique positive periodic solution ωε(t) to Eq (4.6). In addition, ωε(t) is monotonically increasing about ε, and ωε(t)ω(t)>0. We obtain that there exists a positive continuous function ω0(t) such that

    limε0+ωε(t)=ω0(t) in [0,T].

    The uniqueness of the positive solution of Eq (4.6) implies that

    limε0+ωε(t)=ω(t) in [0,T].

    It follows from Lemma 4.1 that

    limε0+θεμ(x,t)=θμ(x,t) in C1,12(ˉQT).

    This means that θ(t) is positive, together with (4.4)–(4.6). Thus, we must have

    θ(t)ω(t) in [0,T].

    This also implies that

    limμθμ(x,t)=ω(t) in C1,12(ˉQT),

    holds for the entire sequence.

    We consider the positive solutions of the periodic-parabolic logistic equation with indefinite weight function and nonhomogeneous diffusion coefficient. When the dispersal rate is small, we can obtain a similar result as in the homogeneous diffusion coefficient. Here we are interested in the case of large diffusion coefficient with nonhomogeneous diffusion coefficient.

    In Theorem 1.1, we obtain the condition of m(x,t) to guarantee a positive periodic solution for all μ>0. Then we investigate the effect of large μ on the positive solution and establish that the limiting profile is determined by the positive solution of Eq (1.5). More precisely, we prove that the positive periodic solution tends to the unique positive solution of the corresponding non-autonomous logistic equation when the diffusion rate is large.

    M. Fan was responsible for writing the original draft. J. Sun handled the review and supervision.

    The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

    The authors would like to thank the anonymous reviewers for the careful reading and several valuable comments to revise the paper. Fan was supported by Gansu postgraduate scientifc research (20230XZX-055) and Sun was supported by NSF of China (12371170) and NSF of Gansu Province of China (21JR7RA535, 21JR7RA537).

    The authors declare there is no conflict of interest.



    [1] S. Javed, M. Danelljan, F. S. Khan, M. H. Khan, M. Felsberg, J. Matas, Visual object tracking with discriminative filters and siamese networks: A survey and outlook, IEEE Trans. Pattern Anal. Mach. Intell., 45 (2023), 6552–6574. https://doi.org/10.1109/TPAMI.2022.3212594 doi: 10.1109/TPAMI.2022.3212594
    [2] F. Chen, X. Wang, Y. Zhao, S. Lv, X. Niu, Visual object tracking: A survey, Comput. Vision Image Understanding, 222 (2022), 103508. https://doi.org/10.1016/j.cviu.2022.103508 doi: 10.1016/j.cviu.2022.103508
    [3] D. Zhang, Z. Zheng, M. Li, R. Liu, CSART: Channel and spatial attention-guided residual learning for real-time object tracking, Neurocomputing, 436 (2021), 260–272. https://doi.org/10.1016/j.neucom.2020.11.046 doi: 10.1016/j.neucom.2020.11.046
    [4] F. Gu, J. Lu, C. Cai, Q. Zhu, Z. Ju, RTSformer: A robust toroidal transformer with spatiotemporal features for visual tracking, IEEE Trans. Hum.-Mach. Syst., 54 (2024), 214–225. https://doi.org/10.1109/THMS.2024.3370582 doi: 10.1109/THMS.2024.3370582
    [5] Y. Qian, L. Yu, W. Liu, A. G. Hauptmann, Electricity: An efficient multi-camera vehicle tracking system for intelligent city, in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), (2020), 2511–2519. https://doi.org/10.1109/CVPRW50498.2020.00302
    [6] X. Chen, X. Xu, Y. Yang, Y. Huang, J. Chen, Y. Yan, Visual ship tracking via a hybrid kernelized correlation filter and anomaly cleansing framework, Appl. Ocean Res., 106 (2021), 102455. https://doi.org/10.1016/j.apor.2020.102455 doi: 10.1016/j.apor.2020.102455
    [7] H. Zhang, Y. Li, H. Liu, D. Yuan, Y. Yang, Feature block-aware correlation filters for real-time UAV tracking, IEEE Signal Process. Lett., 31 (2024), 840–844. https://doi.org/10.1109/LSP.2024.3373528 doi: 10.1109/LSP.2024.3373528
    [8] X. Wang, D. Zeng, Y. Li, M. Zou, Q. Zhao, S. Li, Enhancing UAV tracking: a focus on discriminative representations using contrastive instances, J. R.-Time Image Process., 21 (2024), 78. https://doi.org/10.1007/s11554-024-01456-2 doi: 10.1007/s11554-024-01456-2
    [9] C. Zhu, J. Yang, Z. Shao, C. Liu, Vision based hand gesture recognition using 3D shape context, IEEE/CAA J. Autom. Sin., 8 (2021), 1600–1613. https://doi.org/10.1109/JAS.2019.1911534 doi: 10.1109/JAS.2019.1911534
    [10] M. N. H. Mohd, M. S. M. Asaari, O. L. Ping, B. A. Rosdi, Vision-based hand detection and tracking using fusion of kernelized correlation filter and single-shot detection, Appl. Sci., 13 (2023), 7433. https://doi.org/10.3390/app13137433 doi: 10.3390/app13137433
    [11] J. F. Henriques, R. Caseiro, P. Martins, J. Batista, High-speed tracking with kernelized correlation filters, IEEE Trans. Pattern Anal. Mach. Intell., 37 (2015), 583–596. https://doi.org/10.1109/TPAMI.2014.2345390 doi: 10.1109/TPAMI.2014.2345390
    [12] Y. Li, J. Zhu, A scale adaptive kernel correlation filter tracker with feature integration, in Computer Vision-ECCV 2014 Workshops, 8926 (2014), 254–265. https://doi.org/10.1007/978-3-319-16181-5_18
    [13] M. Danelljan, G. Hager, F. S. Khan, M. Felsberg, Learning spatially regularized correlation filters for visual tracking, in 2015 IEEE International Conference on Computer Vision (ICCV), (2015), 4310–4318. https://doi.org/10.1109/ICCV.2015.490
    [14] C. Ma, X. Yang, C. Zhang, M. Yang, Long-term correlation tracking, in 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (2015), 5388–5396. https://doi.org/10.1109/CVPR.2015.7299177
    [15] M. Danelljan, G. Hä ger, F. S. Khan, M. Felsberg, Discriminative scale space tracking, IEEE Trans. Pattern Anal. Mach. Intell., 39 (2017), 1561–1575. https://doi.org/10.1109/TPAMI.2016.2609928 doi: 10.1109/TPAMI.2016.2609928
    [16] M. Danelljan, G. Bhat, F. Shahbaz Khan, M. Felsberg, ECO: Efficient convolution operators for tracking, in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (2017), 6931–6939. https://doi.org/10.1109/CVPR.2017.733
    [17] A. Lukezic, T. Vojir, L. C. Zajc, J. Matas, M. Kristan, Discriminative correlation filter with channel and spatial reliability, in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (2017), 4847–4856. https://doi.org/10.1109/CVPR.2017.515
    [18] Z. Huang, C. Fu, Y. Li, F. Lin, P. Lu, Learning aberrance repressed correlation filters for real-time UAV tracking, in 2019 IEEE/CVF International Conference on Computer Vision (ICCV), (2019), 2891–2900. https://doi.org/10.1109/ICCV.2019.00298
    [19] B. Li, W. Wu, Q. Wang, F. Zhang, J. Xing, J. Yan, SiamRPN++: Evolution of siamese visual tracking with very deep networks, in 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), (2019), 4277–4286. https://doi.org/10.1109/CVPR.2019.00441
    [20] T. Xu, Z. Feng, X. Wu, J. Kittler, Joint group feature selection and discriminative filter learning for robust visual object tracking, in 2019 IEEE/CVF International Conference on Computer Vision (ICCV), (2019), 7949–7959. https://doi.org/10.1109/ICCV.2019.00804
    [21] D. S. Bolme, J. R. Beveridge, B. A. Draper, Y. M. Lui, Visual object tracking using adaptive correlation filters, in 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, (2010), 2544–2550. https://doi.org/10.1109/CVPR.2010.5539960
    [22] L. Bertinetto, J. Valmadre, J. F. Henriques, A. Vedaldi, P. H. Torr, Fully-convolutional siamese networks for object tracking, in Computer Vision-ECCV 2016 Workshops, 9914 (2016), 850–865. https://doi.org/10.1007/978-3-319-48881-3_56
    [23] H. K. Galoogahi, A. Fagg, S. Lucey, Learning background-aware correlation filters for visual tracking, in 2017 IEEE International Conference on Computer Vision (ICCV), (2017), 1144–1152. https://doi.org/10.1109/ICCV.2017.129
    [24] Y. Li, C. Fu, F. Ding, Z. Huang, G. Lu, Autotrack: Towards high-performance visual tracking for UAV with automatic spatio-temporal regularization, in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), (2020), 11920–11929. https://doi.org/10.1109/CVPR42600.2020.01194
    [25] Z. Song, J. Yu, Y. P. Chen, W. Yang, Transformer tracking with cyclic shifting window attention, in 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), (2022), 8781–8790. https://doi.org/10.1109/CVPR52688.2022.00859
    [26] Y. Chen, H. Wu, Z. Deng, J. Zhang, H. Wang, L. Wang, et al., Deep-feature-based asymmetrical background-aware correlation filter for object tracking, Digital Signal Process., 148 (2024), 104446. https://doi.org/10.1016/j.dsp.2024.104446 doi: 10.1016/j.dsp.2024.104446
    [27] K. Chen, L. Wang, H. Wu, C. Wu, Y. Liao, Y. Chen, et al., Background-aware correlation filter for object tracking with deep CNN features, Eng. Lett., 32 (2024), 1353–1363.
    [28] R. M. Gray, Toeplitz and circulant matrices: A review, Found. Trends Commun. Inf. Theory, 2 (2006), 155–239. http://doi.org/10.1561/0100000006 doi: 10.1561/0100000006
    [29] J. F. Henriques, R. Caseiro, P. Martins, J. Batista, Exploiting the circulant structure of tracking-by-detection with kernels, in Computer Vision-ECCV 2012, (2012), 702–715. https://doi.org/10.1007/978-3-642-33765-9_50
    [30] M. E. Kilmer, C. D. Martin, Factorization strategies for third-order tensors, Linear Algebra Appl., 435 (2011), 641–658. https://doi.org/10.1016/j.laa.2010.09.020 doi: 10.1016/j.laa.2010.09.020
    [31] N. Hao, M. E. Kilmer, K. Braman, R. C. Hoover, Facial recognition using tensor-tensor decompositions, SIAM J. Imaging Sci., 6 (2013), 437–463. https://doi.org/10.1137/110842570 doi: 10.1137/110842570
    [32] M. E. Kilmer, K. Braman, N. Hao, R. C. Hoover, Third-order tensors as operators on matrices: A theoretical and computational framework with applications in imaging, SIAM J. Matrix Anal. Appl., 34 (2013), 148–172. https://doi.org/10.1137/110837711 doi: 10.1137/110837711
    [33] B. Hunt, A matrix theory proof of the discrete convolution theorem, IEEE Trans. Audio Electroacoust., 19 (1971), 285–288. https://doi.org/10.1109/TAU.1971.1162202 doi: 10.1109/TAU.1971.1162202
    [34] J. Martinez, R. Heusdens, R. C. Hendriks, A generalized Fourier domain: Signal processing framework and applications, Signal Process., 93 (2013), 1259–1267. https://doi.org/10.1016/j.sigpro.2012.10.015 doi: 10.1016/j.sigpro.2012.10.015
    [35] A. Iwasaki, Deriving the variance of the discrete Fourier transform test using Parseval's theorem, IEEE Trans. Inf. Theory, 66 (2020), 1164–1170. https://doi.org/10.1109/TIT.2019.2947045 doi: 10.1109/TIT.2019.2947045
    [36] Q. Hu, H. Wu, J. Wu, J. Shen, H. Hu, Y. Chen, et al., Spatio-temporal self-learning object tracking model based on anti-occlusion mechanism, Eng. Lett., 31 (2023), 1–10.
    [37] Y. Huang, Y. Chen, C. Lin, Q. Hu, J. Song, Visual attention learning and antiocclusion-based correlation filter for visual object tracking, J. Electron. Imaging, 32 (2023), 13023. https://doi.org/10.1117/1.JEI.32.1.013023 doi: 10.1117/1.JEI.32.1.013023
    [38] J. Cui, J. Wu, L. Zhao, Learning channel-selective and aberrance repressed correlation filter with memory model for unmanned aerial vehicle object tracking, Front. Neurosci., 16 (2023). https://doi.org/10.3389/fnins.2022.1080521 doi: 10.3389/fnins.2022.1080521
    [39] C. Fan, H. Yu, Y. Huang, C. Shan, L. Wang, C. Li, SiamON: Siamese occlusion-aware network for visual tracking, IEEE Trans. Circuits Syst. Video Technol., 33 (2023), 186–199. https://doi.org/10.1109/TCSVT.2021.3102886 doi: 10.1109/TCSVT.2021.3102886
    [40] W. Hu, Q. Wang, L. Zhang, L. Bertinetto, P. H. S. Torr, SiamMask: A framework for fast online object tracking and segmentation, IEEE Trans. Pattern Anal. Mach. Intell., 45 (2023), 3072–3089.
    [41] D. Sharma, Z. A. Jaffery, Multiple object tracking through background learning, Comput. Syst. Sci. Eng., 44 (2023), 191–204. https://doi.org/10.32604/csse.2023.023728 doi: 10.32604/csse.2023.023728
    [42] J. Zhang, Y. He, S. Wang, Learning adaptive sparse spatially-regularized correlation filters for visual tracking, IEEE Signal Process. Lett., 30 (2023), 11–15. https://doi.org/10.1109/LSP.2023.3238277 doi: 10.1109/LSP.2023.3238277
  • This article has been cited by:

    1. Peng E, Tingting Xu, Linhua Deng, Yulin Shan, Miao Wan, Weihong Zhou, Solutions of a class of higher order variable coefficient homogeneous differential equations, 2025, 20, 1556-1801, 213, 10.3934/nhm.2025011
  • Reader Comments
  • © 2024 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1276) PDF downloads(62) Cited by(2)

Figures and Tables

Figures(7)  /  Tables(5)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog