Research article

Multi-level thresholding image segmentation for rubber tree secant using improved Otsu's method and snake optimizer


  • The main disease that decreases the manufacturing of natural rubber is tapping panel dryness (TPD). To solve this problem faced by a large number of rubber trees, it is recommended to observe TPD images and make early diagnosis. Multi-level thresholding image segmentation can extract regions of interest from TPD images for improving the diagnosis process and increasing the efficiency. In this study, we investigate TPD image properties and enhance Otsu's approach. For a multi-level thresholding problem, we combine the snake optimizer with the improved Otsu's method and propose SO-Otsu. SO-Otsu is compared with five other methods: fruit fly optimization algorithm, sparrow search algorithm, grey wolf optimizer, whale optimization algorithm, Harris hawks optimization and the original Otsu's method. The performance of the SO-Otsu is measured using detail review and indicator reviews. According to experimental findings, SO-Otsu performs better than the competition in terms of running duration, detail effect and degree of fidelity. SO-Otsu is an efficient image segmentation method for TPD images.

    Citation: Shenghan Li, Linlin Ye. Multi-level thresholding image segmentation for rubber tree secant using improved Otsu's method and snake optimizer[J]. Mathematical Biosciences and Engineering, 2023, 20(6): 9645-9669. doi: 10.3934/mbe.2023423

    Related Papers:

    [1] Benjamin H. Singer, Denise E. Kirschner . Influence of backward bifurcation on interpretation of R0 in a model of epidemic tuberculosis with reinfection. Mathematical Biosciences and Engineering, 2004, 1(1): 81-93. doi: 10.3934/mbe.2004.1.81
    [2] Kento Okuwa, Hisashi Inaba, Toshikazu Kuniya . An age-structured epidemic model with boosting and waning of immune status. Mathematical Biosciences and Engineering, 2021, 18(5): 5707-5736. doi: 10.3934/mbe.2021289
    [3] Zhiping Liu, Zhen Jin, Junyuan Yang, Juan Zhang . The backward bifurcation of an age-structured cholera transmission model with saturation incidence. Mathematical Biosciences and Engineering, 2022, 19(12): 12427-12447. doi: 10.3934/mbe.2022580
    [4] Aili Wang, Yanni Xiao, Huaiping Zhu . Dynamics of a Filippov epidemic model with limited hospital beds. Mathematical Biosciences and Engineering, 2018, 15(3): 739-764. doi: 10.3934/mbe.2018033
    [5] Liumei Wu, Baojun Song, Wen Du, Jie Lou . Mathematical modelling and control of echinococcus in Qinghai province, China. Mathematical Biosciences and Engineering, 2013, 10(2): 425-444. doi: 10.3934/mbe.2013.10.425
    [6] Elamin H. Elbasha . Model for hepatitis C virus transmissions. Mathematical Biosciences and Engineering, 2013, 10(4): 1045-1065. doi: 10.3934/mbe.2013.10.1045
    [7] Yuyi Xue, Yanni Xiao . Analysis of a multiscale HIV-1 model coupling within-host viral dynamics and between-host transmission dynamics. Mathematical Biosciences and Engineering, 2020, 17(6): 6720-6736. doi: 10.3934/mbe.2020350
    [8] Muntaser Safan, Bayan Humadi . An SIS sex-structured influenza A model with positive case fatality in an open population with varying size. Mathematical Biosciences and Engineering, 2024, 21(8): 6975-7011. doi: 10.3934/mbe.2024306
    [9] Kento Okuwa, Hisashi Inaba, Toshikazu Kuniya . Mathematical analysis for an age-structured SIRS epidemic model. Mathematical Biosciences and Engineering, 2019, 16(5): 6071-6102. doi: 10.3934/mbe.2019304
    [10] Lin Chen, Xiaotian Wu, Yancong Xu, Libin Rong . Modelling the dynamics of Trypanosoma rangeli and triatomine bug with logistic growth of vector and systemic transmission. Mathematical Biosciences and Engineering, 2022, 19(8): 8452-8478. doi: 10.3934/mbe.2022393
  • The main disease that decreases the manufacturing of natural rubber is tapping panel dryness (TPD). To solve this problem faced by a large number of rubber trees, it is recommended to observe TPD images and make early diagnosis. Multi-level thresholding image segmentation can extract regions of interest from TPD images for improving the diagnosis process and increasing the efficiency. In this study, we investigate TPD image properties and enhance Otsu's approach. For a multi-level thresholding problem, we combine the snake optimizer with the improved Otsu's method and propose SO-Otsu. SO-Otsu is compared with five other methods: fruit fly optimization algorithm, sparrow search algorithm, grey wolf optimizer, whale optimization algorithm, Harris hawks optimization and the original Otsu's method. The performance of the SO-Otsu is measured using detail review and indicator reviews. According to experimental findings, SO-Otsu performs better than the competition in terms of running duration, detail effect and degree of fidelity. SO-Otsu is an efficient image segmentation method for TPD images.



    Mathematical epidemiology constructs mathematical models to understand the dynamics of transmissible diseases, study disease control and elimination, compare strategies of treatments and vaccinations, predict the peak time of an epidemic, and estimate the final size of an outbreak. In the process, mathematical theories on transmission dynamics of infectious diseases are enriched.

    It was Kermack and McKendrick who established the celebrated threshold theory to prognosticate the occurrence of an epidemic [1,2]. Their threshold was given explicitly in terms of the initial size of the susceptible population (the size of the community). If the initial size of the susceptible population is less than the threshold value, no epidemic can occur; otherwise, it can occur. The use of mass-action law by Kermack and McKendrick was the fundamental for them to address the threshold in terms of the density of the susceptible. Adding vital dynamics from Kermack and McKendrick's epidemic models, Hethcote [3,4] showed that the threshold is a global critical value (regardless of the initial conditions) for S-I-S models and S-I-R models (even when vaccination was included). Lajmanovich and York [5] and Hethcote and van Ark [6] found that the global threshold is valid for multi-group S-I-S models for the spread of gonorrhea, but each group had constant sizes.

    Starting in the early 1980's, the threshold for epidemic models has been consistently denoted by R0, which is now formally defined as the expected number of secondary cases produced by a typical infected individual during its entire infectious period in a susceptible population. Though it took many iterations to define R0 [7], each step along the way to properly interpret this critical biological number is a prominent landmark in mathematical epidemiology. For instance, basic reproduction rate had been seen in [8,9,10]; infectious contact number and replacement number were seen in [3,4,11]; the basic reproduction rate later was replaced by basic reproduction ratio such as in [12,13]; and finally basic reproduction number has been uniformly accepted since the earlier 1990's [14].

    As a fundamental threshold, the basic reproduction number plays the foremost role in determining whether a disease can take off or not. If the basic reproduction number for a disease is greater than one, that disease can invade a population. If it's less than one, the disease dies out (for instance see [12,13]). The primary purpose of this paper, however, is to demonstrate that the beauty of R0 is not the only story in mathematical epidemiology.

    In fact, many epidemiological models have questioned whether a single threshold could completely determine the dynamics of a disease. One has observed that multi-threshold values are necessary to completely determine the dynamics of the disease [15,16,17,18]. As a result, a collection of epidemic models have shown that bi-stability occurs when R0<1, which typically has been driven by the occurrence of backward bifurcations [19,20,21,22].

    As is shown schematically in Figure 1, when the bifurcation at R0=1 is forward, R0>1 implies disease can invade population; while R0<1, the disease cannot invade the population. If the bifurcation is backward at R0=1, there exists a stable equilibrium (even a stable limit cycle exists) when R0<1 [17], i.e., the disease does not die out even though R0<1.

    Figure 1.  Left figure is a forward bifurcation diagram, right a backward bifurcation diagram.

    The study of backward bifurcations has a fairly long history. We briefly review the earlier research in mathematical epidemiology on backward bifurcations. Mathematical epidemiology literature first reported the multiple endemic equilibria in mathematical models for the spread of HIV in 1989 by Castillo-Chavez and collaborators [23]. It historically challenged that "thinking that S-I-R epidemic models do not have multiple endemic equilibria is not accurate". Their later papers further confirmed that multiple group model of S-I-R type can have multiple endemic equilibria due to the varying sub-population sizes [24]. By considering that the birth rate is affected by the disease, Diekmann and Kretzschmar observed multiple stable equilibria in 1991 [25]. However, neither [23] nor [25] explicitly used the words– backward bifurcation. It wasn't until 1995 that Hadeler and Castillo-Chavez first formally used the name: backward bifurcation, and its the occurrence was due to the introduction of core group and non-core group [26].

    Hadeler and van der Driessche's paper in 1997 entitled "Backward bifurcation in epidemic control" [22] studied groups of two that have different susceptibilities can lead to the bi-stability situation. Dushoff et al. explored possible mechanisms behind backward bifurcations in simple disease models in 1998 [27]. This research further developed a theoretical local bifurcation indicator that determines whether forward or backward from earlier work in [24]. Later, this indicator was well formulated in [18] and [21]. The new millennium saw an new exploration in the study of backward bifurcations. Variety of mechanisms behind backward bifurcations have been identified in epidemic models. Castillo-Chavez and Feng [15] constructed dynamic model for the transmission of tuberculosis, claiming that exogenous reinfections can result in backward bifurcations, which was later found in other mathematical models for the transmission of tuberculosis [28,29,30]. It was shown by Kribs-Zaleta and Velasco-Hernández in [31] and by Brauer in [19,20] that vaccinations can trigger backward bifurcation. Martcheva and Thieme showed that super-infections, a sort of reinfection, are one mechanism in generating backward bifurcations [32]. When looking back the mathematical models for tuberculosis [15,28,29,30] collectively, one can conclude that reinfection would naturally cause backward bifurcations. Wang [33] and our previous work [17] showed that non-standard treatment strategies can cause the occurrence of backward bifurcation. Here, we very briefly review some well-known studies for backward bifurcations in deterministic differential equation models for disease transmission. It is worth mentioning that backward bifurcations have been found in stochastic and discrete epidemic models (for instance, see [34,35]).

    The basic reproduction number was found in measuring the primary infection force. When reinfections take place, how do we measure the reinfection forces? Corresponding to R0, this paper finds a basic reinfection number that measures the reinfection forces, which turns out to be a new tipping point for disease dynamics.

    The following Theorem 1 is a successful and convenient approach in determining whether or not a system exhibits a backward bifurcation ([18,21,27]). It will be repeatedly used in this article to guide all our calculations.

    Theorem 1. [21] Consider a system of ordinary differential equations

    dxdt=f(x,ϕ),f:Rn×RRn  and  fC2(Rn×R), (1.1)

    with a parameter ϕ. Assume that:

    1. 0 is an equilibrium of the system, that is, f(0,ϕ)0 for all ϕ; and

    2. Zero is a simple eigenvalue of Dxf(0,0)=(fixj(0,0)) and all other eigenvalues of Dxf(0,0) have negative real parts.

    Let W=[w1,w2,,wn] and V=[v1,v2,,vn] be a right and a left eigenvector matrix Dxf(0,0), respectively, corresponding to the zero eigenvalue; and let fk(x,ϕ) be the kth component of f(x,ϕ). Then the local dynamics of system (1.1) around the equilibrium 0 is totally determined by the signs of a and b below:

    a=nk,i,j=1vkwiwj2fkxixj(0,0), (1.2)
    b=nk,i=1vkwi2fkxiϕ(0,0). (1.3)

    Particularly, if b>0 and a>0, then a backward bifurcation occurs for system (1.1) at ϕ=0; while b>0 and a<0, a forward bifurcation occurs.

    When applying Theorem 1 to an epidemiological model, the equilibrium 0 corresponds to the disease-free equilibrium and ϕ=R01. That is, ϕ=0 corresponds to R0=1. Theorem 1 here is not exactly the same as Thorem 4.1 in [21], because we have dropped the assumption that W is nonnegative (see Remark 1 in [21]).

    In this section, we use a simple epidemic model to introduce our essential concept: the basic reinfection number.

    We first introduce a mathematical model for a socially transmitted disease such as cigarettes smoking (or drug abuse), where peer pressure or peer influence presumably plays a critical role. The entire population is divided into nonsmokers, smokers, temporary quitters, and permanent quitters. Let S be the number of nonsmokers; I the number of smokers; R the number of the individuals who temporarily quit smoking but have not decided if quit smoking forever; and T the number of individuals who has commit to quitting cigarettes smoking permanently. A system of ordinary differential equations is used to model the dynamics of cigarettes smoking. The model equations read

    {dSdt=μNβ1SINμS,dIdt=β1SIN+β2RINμIαI,dRdt=αIβ2RINμRγR,dTdt=γRμT,N=S+I+R+T. (2.1)

    The per-capita birth rate and the per-capita death rate are assumed to equal (μ). β1SIN is the rate of new smokers due to peer pressure. β2RIN is the relapse rate, which is also driven by peer pressure. αI is the rate of temporary quitting. γR is the rate of permanent quitting.

    In the study of epidemics models, it has become well known that the infection force β1SIN is measured by the basic reproduction number. That is the well-known threshold phenomenon. When R0>1, the disease takes off; and R0<1 the disease dies out. Now we are interested in how to measure the reinfection force (relapse rate) β2RIN and the role of the reinfection force in determining the dynamics of the disease.

    The basic reproduction number for model (2.1) is given by

    R0=β1μ+α. (2.2)

    But the dynamics of the model (2.1) can be described by two quantities: R0 and

    Rr=(β2μ+α)(αμ+α)(μμ+γ). (2.3)

    Theorem 2. Consider model (2.1). When Rr<1, a forward bifurcation occurs at R0=1; Conversely, when Rr>1, a backward bifurcation occurs at R0=1.

    According to Theorem 2, the direction of bifurcation at R0=1 is completely determined by Rr. We define Rr=(β2μ+α)(αμ+α)(μμ+γ) to be the basic reinfection number for the SIRT model (2.1). Formally, basic reinfection number, denoted by Rr, is defined to be the average number of reinfections of a typical infectious during entire infectious period in a subpopulation of all recovered individuals.

    The basic reinfection number Rr in (2.3) has clear epidemiological and biological meaning. Exactly the same as R0, Rr measures both the availability of resources and capability to reinforce. Each term in the expression has clear demographic or epidemiological implication.

    Obviously, β2 is the transmission rate of reinfection; 1μ+α is the average reinfection period which is the same as the average infection period. The first part of the expression in (2.3) is the same as R0, except the for reinfection rate. The second and the third parts imply that the reinfection takes place in the recovered subpopulation (R). To become a member of the R-class, an infected individual must recover from I-class to R-class. This occurs with the probability α/(α+μ). Furthermore, an individual in R-class can be reinfected only if the individual does not become a member of T-class. This event happens with the probability μ/(γ+μ). Overall (α/(α+μ))(μ/(γ+μ)) gives the probability of contacting the recovered class. Comparing the expressions (2.3) and (2.2), it can be seen that the only difference is that they each apply to distinct sub-populations. The basic reproduction number is applied to the general susceptible population (naive population). The probability of contacting the susceptible class is one because it is assumed that all individuals are susceptible from the definition of R0. However, this is not the case for basic reinfection number, since reinfections take place only if the recovered population is established. The term (α/(α+μ))(μ/(γ+μ)) in (2.3) is exact the probability of the establishment of the recovered population. Therefore, the basic reinfection number is applied to the recovered population. In short, the basic reproduction number takes care of the primary infections and the basic reinfection number accounts for reinfections, which were missing in R0 (It is beyond the reach of R0).

    Below, we mathematically prove Theorem 2.

    Proof. The Jacobian of system (2.1) around the smoking-free equilibrium (N,0,0,0) is given by

    J DFE =(μβ1000β1(α+μ)000α(μ+γ)000γμ).

    Choosing β1 as the bifurcation parameter, then R0=1 corresponds to β1=β1=μ+α. It can be seen that λ=0 is a simple eigenvalue of the matrix J(DFE,β=β1). A right and left eigenvector corresponding to the zero eigenvalue are W=[w1,w2,w3,w4]=[β1μ,1,αμ+γ,αγμ(μ+γ)] and V=[v1,v2,v3,v4]=[0;1;0;0], respectively. Using formula of (1.2) and (1.3), we obtain b=Λμ>0 and a=2β1w1w2+2β2αμ+γ which can be conveniently expressed as a2=β2αμ+γ(μ+α)2μ. Therefore, a backward bifurcation occurs if β2αμ+γ(μ+α)2μ>0. It is equivalent to (β2μ+α)(αμ+α)(μμ+γ)>1, i.e. Rr>1. Applying Theorem 1, Theorem 2 follows.

    This section demonstrates the basic reinfection numbers in some epidemic models for tuberculosis and socially transmitted diseases.

    For more general tuberculosis (TB) exogenous reinfection models, we refer the reader to [15,21]. Here, we construct a simpler model to illustrate the infection number. Basically, we assume that successfully treated individuals are still susceptible individuals. The host population is divided into the following three epidemiological classes: susceptible (S), exposed (L, infected but not infectious), infectious (I), and successfully treated individuals are still susceptible. Let N denote the total population size. The model takes the following ODE form of three equations:

    {dSdt=Λβ1SINμS+r2L+r1I,dLdt=β1SINβ2LIN(μ+k+r2)L,dIdt=β2LIN+kL(μ+d+r1)I. (3.1)

    Here, Λ is the recruitment rate of the population; β1 is the average number of effective contacts a susceptible has per unit of time; β2 is the average number of effective contacts an exposed individual has per unit of time. μ is the per-capita natural death rate; k is the primary progression rate at which an exposed individual leaves the latent class by becoming a member of an infectious class; d is the per-capita disease-induced death rate; r2 and r1 are the per-capita treatment rates for the exposed and the infected, respectively. The term β2LIN models the exogenous reinfection rates. The treated infections and exposed become members of susceptible class. The basic reproduction number for (3.1) is

    R0=(β1μ+d+r1)(kμ+k+r2).

    Because of the exogenous reinfection, model (3.1) undergoes a richer dynamics due to the occurrence of backward bifurcation, which can be characterized by the basic reinfection number.

    The basic reinfection number for model (3.1) is

    Rr=(β2μ+d+r1+k)(μ+r2μ+k+r2). (3.2)

    Again, we conclude that the basic reinfection number characterizes the occurrence of backward bifurcation. We formally summarize this result into theorem 3 below. The proof of this theorem can be found in Appendix A.

    Theorem 3. Consider model (3.1). The basic reinfection number is given by equation (3.2). When Rr<1, a forward bifurcation occurs at R0=1; Conversely, when Rr>1, a backward bifurcation occurs at R0=1.

    Similar to the interpretation to the basic reinfection number for SIRT model (2.1), here is an explanation to the basic infection number for model (3.1). μ+r2μ+k+r2 is the fraction of exposed population that does not naturally progress to active TB. This is the target population for exogenous reinfections. Only this portion of exposed population can be reinfected. It was observed that 1μ+d+r1+k tends to be the average reinfection period, which is shorter than the average primary infection period 1μ+d+r1.

    A perfect match of basic reproduction number and the basic reinfection number can be identified in an extreme case where k=0. In this case, R0=0 and there is no primary progression and all new cases are the results of the exogenous reinfections. It is amazing that Rr=β2μ+d+r1 has the exactly same formula as the basic reproduction number. If this number is greater than one, then there is possibility that the disease becomes endemic. If the number is less than one, the disease dies out. This study has observed an incredible result: a disease can become endemic even though the basic reproduction number is equal to zero provided that there are reinfections and basic reinfection number is bigger then one.

    Though regular mycobacterium tuberculosis is curable, it takes eight months to clear the bacterium with a complete treatment regime of antibiotics. Incomplete treatment puts the patient at higher risk of developing antibiotic resistance. Now we extend our model in previous subsection to consider multiple strains of tuberculosis infections. An antibiotic sensitive strain of TB and an antibiotic resistant strain are considered. Our goal is to naturally extend our basic reinfection number from one strain to multiple strains.

    Notations are the same as model (3.1). Here, we simply add subscripts to distinguish the different strains. The first strain is drug-sensitive and the second is drug-resistant. Let L1 and L2 denote individuals exposed to regular tuberculosis (antibiotic sensitive) and antibiotic resistant tuberculosis, respectively. Likewise, I1 and I2 stand for infectious individuals of antibiotic sensitive and antibiotic resistant, respectively. Treatments are applied to both the exposed and the infectious of the sensitive strain. p is the proportion of exposed individuals of antibiotic sensitive who did not complete their treatment and develop into antibiotic resistance. 1p is the proportion of successfully treated exposed individuals. Hence, the treatment rate r2L1 for the exposed class is broken into successful treatment rate (1p)r2L1 and unsuccessful treatment rate pr2L1. The former becomes member of S class, the latter goes to L2 class. Let q be the proportion of infectious individuals of antibiotic sensitive who did not complete their treatment and develop into antibiotic resistant; 1q be the proportion of successful treatment for infectious individuals. Among total treatment rate of r1I1, qr1I1 gives the rate at which individuals develop resistant tuberculosis; and (1q)r1I1 goes to S class as the result of successful treatment. The model takes the following form of ordinary differential equations:

    {dSdt=Λβ1(I1+I2)NSμS+(1p)r2L1+(1q)r1I1,dL1dt=β1I1NSβ2I1NL1(μ+k1+r2)L1,dL2dt=β1I2NS+pr2L1+qr1I1β2I2NL2(μ+k2)L2,dI1dt=β2I1NL1+k1L1(μ+d1+r1)I1,dI2dt=β2I2NL2+k2L2(μ+d2)I2. (3.3)

    Computations in Appendix D show that R10=(β1μ+d1+r1)(k1μ+k1+r2) is the basic reproduction number for drug-sensitive strain and R20=(β1μ+d2)(k2μ+k2) is the basic reproduction number for drug-resistant strain. Then the basic reproduction number for whole system is

    R0=max{(β1μ+d1+r1)(k1μ+k1+r2),(β1μ+d2)(k2μ+k2)}. (3.4)

    In Appendix D, a detailed computation for formula (3.4) is presented.

    The situation is similar to that of the basic reproduction number (3.4). For each strain of TB, there is a corresponding basic reinfection number. The basic reinfection number for sensitive strain is

    R1r=(β2μ+d1+r1+k1)(μ+r2μ+r2+k1),

    which is derived from Theorem 1 by assuming R10>R20. On the other hand, if R10<R20, the basic reinfection number for resistant strain of TB is

    R2r=(β2μ+d2+k2)(μμ+k2).

    Hence, the basic reinfection number for the entire system is

    Rr={(β2μ+d1+r1+k1)(μ+r2μ+r2+k1), if  R10>R20(β2μ+d2+k2)(μμ+k2), if   R10<R20 (3.5)

    Theorem 4. Consider model (3.3). The basic reinfection number is given by (3.5). When Rr<1, a forward bifurcation occurs at R0=1; Conversely, when Rr>1, a backward bifurcation occurs at R0=1.

    We conclude again that the basic reinfection number controls the occurrence of backward bifurcation. Biological interpretation for the basic infection number is exactly the same as (3.2) of TB model (3.1) for a single strain. Both resistant strain and sensitive strain include the exogenous reinfection which causes the occurrence of backward bifurcations. What if both strains generate bi-stability? Numerically, tri-stability dynamics were found in [28]. That is, a collision of two backward bifurcations is able to generate tri-stability dynamics. We refer the interested readers to the examples in [28].

    Children who smoked once before age 11 are twice as likely to become a regular smoker by age 14 [37]. This is so called the sleeper effects. In this subsection, we present a mathematical model to study the dynamics of smoking in the United States among children ages 11 to 18 [38].

    All children enter the system as susceptible non-smokers with either little or no smoking history. Consequently, a two-string SIRT model is constructed, where string 1 is composed of individuals who smoked once before age 11, and string 2 is composed of those who did not. Using this model, the "sleeper effect" can be investigated.

    Each string is composed of four classes; Ni, Si, Qi (where i=1,2 denotes the respective string), and P. The susceptible non-smoker classes (Ni) are composed of individuals who have never been regular smokers. The regular smoker classes (Si) are composed of smokers. The recovered classes (Qi) are composed of individuals who have quit smoking but are susceptible to relapse. The class of permanent non-smokers (P) is composed of individuals who have decided to permanently quit (from Qi).

    Individuals who turn 11 years old enter either N1 or N2 with rates μrT and μ(1r)T, respectively, where T is the total population of adolescents. Non-smokers (Ni) are susceptible to becoming regular smokers as a result of interactions with regular smokers (Si). Smokers (Si) can quit smoking due to personal choice or education. Individuals who have quit (Qi) can relapse due to social interactions with smokers. All individuals who are not currently regular smokers (Ni, Qi) can move to class P by choosing to permanently abstain from cigarettes. We assume the total population, T, is constant. Definitions of all model parameters are shown in Table 1. The following mathematical model was proposed in [38].

    {dN1dt=rμTβ1N1S1+S2TμN1,dN2dt=(1r)μTβ2N2S1+S2TμN2,dS1dt=β1N1S1+S2T+ψ1Q1S1+S2T(ϕ1+μ)S1,dS2dt=β2N2S1+S2T+ψ2Q2S1+S2T(ϕ2+μ)S2,dQ1dt=ϕ1S1ψ1Q1S1+S2T(α1+μ)Q1,dQ2dt=ϕ2S2ψ2Q2S1+S2T(α2+μ)Q2,dPdt=α1Q1+α2Q2μP,T=N1+N2+Q1+Q2+S1+S2+P. (3.6)
    Table 1.  Parameter Definitions.
    Parameter Definition
    μ The rate of leaving a class as a result of aging or death
    It is also the recruitment rate
    r Fraction of adolescents who smoked before age 11
    βi Conversion rate of new smokers
    ϕi Rate at which individuals quit smoking
    ψi Relapse rate
    αi Rate at which individuals permanently quit smoking

     | Show Table
    DownLoad: CSV

    The equations of N1, S1 and Q1 in model (3.6) describe the dynamics of the population that smoked at least once before age 11. Likewise, the equations of N2, S2 and Q2 describe the dynamics of the population that did not smoke before age 11.

    The basic reproduction number for model (3.6) is

    R0=rβ1ϕ1+μ+(1r)β2ϕ2+μ. (3.7)

    β1ϕ1+μ is the basic reproduction number of subpopulation with a smoking history and β2ϕ2+μ is the basic reproduction numbers of subpopulation without a smoking history. The overall basic reproduction number is the weighted average of the basic reproduction numbers for these two populations using r and (1r) as the weights.

    The general structure of model (3.6) is almost the same as model (2.1). Since each string of (3.6) has the same structure as (2.1), we then expect to observe the same structure of basic reinfection number. To highlight the role of the basic reinfection number in this research, we here consider a simpler case where that β1=β2.

    The basic reinfection number for model (3.6) is

    Rr=rψ1ϕ1+μϕ1ϕ1+μμα1+μ+(1r)ψ2ϕ2+μϕ2α2+μμα2+μ, (3.8)

    where (ψ1ϕ1+μ)(ϕ1ϕ1+μ)(μα1+μ) is the basic reinfection number of subpopulation with a smoking history before age 11. Likewise, (ψ2ϕ2+μ)(ϕ2α2+μ)(μα2+μ) is the basic reinfection number of subpopulation without a smoking history before age 11.

    To interpret the meaning of ψ1ϕ1+μϕ1ϕ1+μμα1+μ, we break it into two components. ψ1ϕ1+μ is the average of relapse due to a typical smoker with smoking history before age 11 since ψ1 is the relapse rate and 1ϕ1+μ is the average smoking period in S1. The second part ϕ1ϕ1+μμα1+μ finds the proportion of potential relapse population, because ϕϕ1+μ is the proportion of quitting in S1, and μα1+μ is the proportion of not permanent abstaining from cigarettes smoking of Q1 class.

    The overall basic reinfection number (3.8) is the weighted average of the basic reinfection numbers of two subpopulations using r and (1r) as the weights. It has the same format as the basic reproduction number (3.7).

    We have already achieved the same conclusion that if Rr>1, the model (3.6) undergoes a backward bifurcation at R0=1. Such a statement is justified rigorously in Appendix C.

    If relapse does not come into play, i.e., ψi=0, then Rr=0<1 and consequently backward bifurcation cannot happen. Namely, Rr>1 happens uniquely because of the relapse. To handle with cigarette smoking in adolescence, relapse must be avoided by all means.

    The reinfection forces (or capability of relapse) is successfully measured by the basic reinfection numbers, the identical role that basic reproduction number plays for the primary infection forces. A primary infection invades a population if the basic reproduction number is greater than one. If reinfection or relapse exist, the basic reinfection indicates a successful invasion of a disease into a population even when the basic reproduction number is less than one. Not only does the basic reinfection number Rr quantify the reinfection forces, but also specifies the type of bifurcation at the basic reproduction number R0=1.

    In reconsidering the remarkable observation from model (3.1) for tuberculosis, where the disease is still able to spread when the basic reproduction number is equal to zero, we may want to combine R0 and Rr together for a complete elimination of the disease. For instance, max{R0,Rr} would be a more proper control number, since the disease dies out when max{R0,Rr}<1 regardless the initial epidemiological status. On one hand, max{R0,Rr} can be applied to epidemic models without reinfections because, when there is no reinfection, the basic reinfection number Rr=0, subsequently max{R0,Rr}=R0. On the other hand, more important observation is that max{R0,Rr} is able to explain why an infectious disease can spread when R0=0. Therefore, as an indicator whether the disease can take off or not, max{R0,Rr} is better than R0 alone.

    Using a few of deterministic epidemic models, we intend to introduce a new concept in theoretic epidemiology. The basic reinfection number Rr has successfully characterized the occurrence of backward bifurcations for a large number of epidemic models. But we found that it is unsuccessful for the appearance of backward bifurcations caused by vaccination factor or nonlinear treatment rate[17,31,33]. Further research would find a corresponding counterpart for the epidemiological model where vaccination matters or nonlinear treatments are applied.

    Unlike R0, which has been well formulated by using the next generation operator [13,18,36], in this paper we do not have a universal formula for Rr. Our approach to formulate it is to compute a and b from Theorem 1. Then manage the inequality a>0 in order to find the basic reinfection number. Future work may be devoted to formulate Rr in general scenarios.

    In the range of R0<1, if reinfection force is strong enough to make Rr>1, the disease may be persistent even when the primary infection cannot support an endemic. This does not work for the full range of R0<1, because when the basic reproduction number is too small, there are not enough recovered individuals to be reinfected. That is why any backward bifurcation curve eventually has to turn around, as can be seen in Figure 4.

    Figure 2.  Diagram of model for cigarette smoking. Relapse of cigarette smoking is highlighted.
    Figure 3.  Diagram of tuberculosis transmission (3.1).
    Figure 4.  A backward bifurcation of system (3.1). This figure is generated by letting β1 change from 0.2 to 0.45, and β2=5, μ=0.15, k=0.1, r1=r2=0.01, d=0. The corresponding basic reinfection number Rr=11.83>1.
    Figure 5.  Diagram of TB model (3.3) of two strains.
    Figure 6.  Diagram of model for cigarettes smoking among young adults. Two-strain model is adopted to take consideration of sleeper effects.

    The author declares that there is no conflict of interest regarding the publication of this paper.

    Here we show how to get the reinfection number for model (3.1). The Jacobian of the system (3.1) around the TB-free equilibrium is

    J=[μr2r1β10(μ+k+r2)β10k(μ+d+r1)].

    When R0=(β1/(μ+d+r1))(k/(μ+k+r2))=1, a left eigenvector of J is V=[v1v2v3]=[0v2v2(μ+k+r2)/k] and a right eigenvector is W=[w1,w2,w3] =[w2(r2+(μ+k+r2)(r1β1)/β1)/μ,w2,w2(μ+k+r2)/β1]. Choosing v2=w2+1 with any w2>0, we obtain that

    b=v2w2>0,anda=v23i,j=1wiwj2f2xixj+v33i,j=1wiwj2f3xixj=2v2w2w3(β1+β2)μ/Λ2v2w3w3β1μ/Λ+2v3w2w3β2μ/Λ.

    To single out the sign of a, we rewrite the last equation as

    a(kΛ/(2v3w2w3μ))=β2(μ+r2)(μ+k+r2)(μ+d+r1+k)

    Obviously, a>0 if and only if (β2μ+d+r1+k)(μ+r2μ+k+r2)>1. Then

    Rr=(β2μ+d+r1+k)(μ+r2μ+k+r2)

    is the basic reinfection number. Applying Theorem 1, we conclude our proof.

    We choose β1 as the bifurcation parameter. Denote the right-hand functions of (3.3) by f1, f2, f3, f4, f5; and the state variables (S,L1,L2,I1,I2) are denoted by (x1,x2,x3,x4,x5). Set K=Λμ for a shortcut.

    Applying theorem 1, we directly compute a and b. The Jacobian of the system evaluated at the TB-free equilibrium (Λ/μ,0,0,0,0) is

    J=[μ(1p)r20(1q)r1β1β10(μ+k1+r2)0β100pr2(μ+k2)qr1β10k10(μ+d1+r1)000k20(μ+d2)]. (12.1)

    Comparing with one-strain TB model (3.1), we here have to additionally deal with whether R0=R10=(β1μ+d1+r1)(k1μ+k1+r2) or R0=R20=(β1μ+d2)(k2μ+k2) because the overall basic reproduction number is R0=max{R10,R20}. For convenience, we tabulate the non-zero second derivatives (2fk/xixj, i,j,k=15) evaluated at the TB-free equilibrium (Λ/μ,0,0,0,0) and R0=1 in Table 2.

    Table 2.  non-zero second derivatives.
    2f2x2x4=β1+β2K 2f2x3x4=β1K 2f2x24=2β1K 2f2x4x5=β1K 2f4x2x4=β2K
    2f3x2x5=β1K 2f3x3x5=β1+β2K 2f3x4x5=β1K 2f3x25=2β1K 2f5x3x5=β2K

     | Show Table
    DownLoad: CSV

    If R10>R20, then R0=1 means R10=1. A left eigenvector of (B.1) at R10=1 is V=[v1,v2,v3,v4,v5]=[0,k1μ+r2+k1v4,0,v4,0], where v4 is free. A right eigenvector is W=[w1,w2,w3,w4,w5], where

    w1=w1(no impact),w2=μ+d1+r1k1w4,w3=k2(pr2(μ+d1+r1)+qr1k1)k1(μ+k2)(μ+d2)k2(μ+d1+r1)(μ+k1+r2),w4=free,w5=(μ+d2)(pr2(μ+d1+r1)+qr1k1)k1(μ+k2)(μ+d2)k2(μ+d1+r1)(μ+k1+r2).

    It can be checked b=k1μ+r2+k1v4w4>0. To compute a, we need to pay attention to f2 and f4 only because v1=v3=v5=0.

    a=v25i,j=1wiwj2f2xixj+v4ni,j=1wiwj2f4xixj

    Looking up Table 2, we obtain that

    a=2v2(w2w42f2x2x4+w242f2x24+w3w42f2x3x4+w5w42f2x5x4)+2v4w2w42f4x2x4a(K2v2w2w4)=(β1+β2)β1w4w2+β2v4v2=(β1+β2)β1k1μ+d1+r1+β2μ+k1+r2k1=β2μ+r2k1(μ+d1+r1+k1)(μ+k1+r2)k1

    Therefore, a>0 iff β2(μ+r2)>(μ+d1+r1+k1)(μ+k1+r2), i.e. β2μ+r1+d1+k1μ+r2μ+k1+r2>1. Then the basic reinfection number for sensitive strain is

    R1r=(β2μ+d1+r1+k1)(μ+r2μ+k1+r2).

    Now assume R10<R20. Then R0=1 means R20=1. A left eigenvector of (12.1) at R20=1 is V=[v1,v2,v3,v4,v5], where v1=0, v2=pr2(μ+d1+r1)+k1qr1k2(μ+d1+r1)(μ+k1+r2)k1(μ+d2)(μ+k2)k22μ+d2v5, v3=k2μ+d2v5, v4=k2qr1(μ+k1+r2)+pr2(μ+d2)(μ+k2)k2(μ+d1+r1)(μ+k1+r2)k1(μ+d2)(μ+k2)k2μ+d2v5, and v5 is free. A right eigenvector is W=[w1,w2,w3,w4,w5]=[(μ+k2)(μ+d2)μk2w5,0,μ+d2k2w5,0,w5]. w5 is free. w2=w4=0 makes the computation of a a lot easier.

    a=v25i,j=1wiwj2f2xixj+v3ni,j=1wiwj2f4xixjv45i,j=1wiwj2f2xixj+v5ni,j=1wiwj2f4xixj=v3(2w2w52f3x2x5+2w3w52f3x3x5+2w4w52f3x4x5+w252f3x25)+2v5w3w52f5x3x5=v3(2w3w52f3x3x5+w252f3x25)+2v5w3w52f5x3x5
    av3w3w5=(22f3x3x5+w5w32f3x25)+2v5v32f5x3x5=2β1+β2K2k2μ+d2β1K+μ+k2k2β2KaK2v3w3w5=β1β2k2μ+d2β1+μ+k2k2β2=β1μ+d2+k2μ+d2+μβ2k2=(μ+k2)(μ+d2+k2)k2+μβ2k2

    Therefore, a>0 iff μβ2>(μ+k2)(μ+d2+k2), i.e. β2μ+d2+k2μμ+k2>1. Then the basic reinfection number for resistant strain is

    R2r=(β2μ+d2+k2)(μμ+k2).

    The basic reinfection number for entire system is

    Rr={(β2μ+d2+k2)(μμ+k2), if   R10>R20(β2μ+d1+r1+k1)(μ+r2μ+r2+k1), if   R10<R20

    The basic reinfection number for model (3.6) is derived in a simple case β1=β2=β. We look at the Jacobian of the system evaluated at the smoker-free equilibrium (rT,(1r)T,0,0,0,0,0)

    J=[μ0βrβr0000μβ(1r)β(1r)00000βr(ϕ1+μ)βr00000β(1r)β(1r)(ϕ2+μ)00000ϕ10(α1+μ)00000ϕ20(α2+μ)00000α1α2μ].

    Choose β as the bifurcation parameter. Apparently R0=1 corresponds to β=β=(ϕ1+μ)(ϕ2+μ)r(ϕ2+μ)+(1r)(ϕ1+μ). A right eigenvector associated with the eigenvalue λ=0 of J is W=[w1,w2,w3,w4,w5,w6,w7], where

    w1=r(ϕ2+μ)(1r)μw4,w2=ϕ2+μμw4,w3=r(ϕ2+μ)(1r)(ϕ1+μ)w4>0,w4=(1+r(ϕ2+μ)(1r)(ϕ1+μ)(ϕ2+μ)(ϕ1+μ))1>0,w5=ϕ1α1+μr(ϕ2+μ)(1r)(ϕ1+μ)w4,w6=ϕ2α2+μw4,w7=α1μϕ1α1+μr(ϕ2+μ)(1r)(ϕ1+μ)w4+α2μϕ2α2+μw4.

    A left eigenvector associated with the eigenvalue λ=0 of J is V=[v1,v2,v3,v4,v5,v6,v7]=[0,0,ϕ2+μϕ1+μ,1,0,0,0]. Without confusion, we set fi to be the right-hand side functions of system (3.6) and denote [N1,N2,S1,S2,Q1,Q2,P] by [x1,x2,x3,x4,x5,x6,x7]. Because of v1=v2=v5=v6=v7=0, only f3 and f4 matter when calculating a and b.

    Applying Theorem 1 to model (3.6), we obtain that

    b=v37i,j=1wiwj2f3xiβ+v47i,j=1wiwj2f4xiβ=rv3(w3+w4)+(1r)v4(w3+w4)=(r(ϕ2+μ)+(1r)(ϕ1+μ))2r(ϕ2+μ)2+(1r)(ϕ1+μ)2>0.
    a=v37i,j=1wiwj2f3xixj+v47i,j=1wiwj2f4xixj=v3[2w1w32f3x1x3+2w1w42f3x1x4+2w3w52f3x3x5+2w4w52f3x4x5]+v4[2w2w32f4x2x3+2w2w42f4x2x4+2w3w62f4x3x6+2w4w62f4x4x6]=2(w3+w4)T(v3βw1+v3ψ1w5+βw2+ψ2w6)

    Organize the above expression to check the sign of a. Noticing w3>0 and w4>0.

    aT2(w3+w4)w4=β[r(ϕ2+μ)(1r)μϕ2+μϕ1+μϕ2+μμ]+ψ1ϕ1α1+μr(ϕ2+μ)(1r)(ϕ1+μ)+ψ2ϕ2α2+μ=βϕ2+μμ[r(1r)ϕ2+μϕ1+μ+1]+ψ1ϕ1α1+μr(ϕ2+μ)(1r)(ϕ1+μ)+ψ2ϕ2α2+μ=(ϕ2+μ)2μ(1r)+ψ1ϕ1α1+μr(ϕ2+μ)(1r)(ϕ1+μ)+ψ2ϕ2α2+μ

    It can be seen that a>0 if and only if

    ψ1ϕ1α1+μr(ϕ2+μ)(1r)(ϕ1+μ)+ψ2ϕ2α2+μ>(ϕ2+μ)2μ(1r),

    which is equivalent to

    rψ1ϕ1+μϕ1ϕ1+μμα1+μ+(1r)ψ2ϕ2+μϕ2α2+μμα2+μ>1.

    Consequently, the basic reinfection number for model (3.6) is

    Rr=r(ψ1ϕ1+μ)(ϕ1ϕ1+μ)(μα1+μ)+(1r)(ψ2ϕ2+μ)(ϕ2α2+μ)(μα2+μ).

    The approach of next-generation operator [18] is used to calculate the basic reproduction number R0 for model (3.3).

    X=[L1,L2,I1,I2], dXdt=F(X)V(X), where

    F(X)=[β1I1KNβ2I2KN00] and V(X)=[β2I1NL1+(μ+k1+r2)L1pr2L1qr1I1+β2I2NL2+(μ+k2)L2β2I1NL1k1L1+(μ+d1+r1)I1β2I2NL2k2L2+(μ+d2)I2].

    F=FX|X=0=[0β00] with β=[β100β1], and V=VX|X=0=[v11v12v21v22], where v11=[μ+k1+r20pr2μ+k2], v12=[00qr10], v21=[k100k2], and v22=[μ+d1+r100μ+d2]. A block format of V1=[γ11γ12γ21γ22]. Since FV1=[βγ21βγ2200], the spectrum radius of FV1 is determined by βγ21. Using block operations of matrices, one can find

    γ21=[k1(μ+d1+r1)(μ+k1+r2)0k1(pr2+k1qr1μ+d1+r1)(μ+d1+r1)(μ+k1+r2)(μ+k2)k2(μ+d2)(μ+k2)]. βγ21=[β1k1(μ+d1+r1)(μ+k1+r2)0β1k1(pr2+k1qr1μ+d1+r1)(μ+d1+r1)(μ+k1+r2)(μ+k2)β1k2(μ+d2)(μ+k2)]. Therefore

    R0=ρ(FV1)=max{(β1μ+d1+r1)(k1μ+k1+r2),(β1μ+d2)(k2μ+k2)},

    and formula (3.4) is proved.



    [1] C. Nayanakantha, Tapping panel dryness: The killer affecting the productivity of rubber plantations, 19 (2021), 26–29.
    [2] R. Putranto, E. Herlinawati, M. Rio, J. Leclercq, P. Piyatrakul, E. Gohet, et al., Involvement of ethylene in the latex metabolism and tapping panel dryness of hevea brasiliensis, Int. J. Mol. Sci., 16 (2015), 17885–17908. https://doi.org/10.3390/ijms160817885 doi: 10.3390/ijms160817885
    [3] Z. Sun, J. Xing, H. Hun, X. Zhang, X. Dong, Y. Deng, Research on recognition and planning of tapping trajectory of natural rubber tree based on machine vision, J. Chin. Agric. Machanization, 43 (2022), 102–108. https://doi.org/10.13733/j.jcam.issn.20955553.2022.05.015 doi: 10.13733/j.jcam.issn.20955553.2022.05.015
    [4] J. Zhang, Y. Liu, H. Xing, Application of improved 2-d entropy algorithm in rubber tree image segmentation, in 2019 2nd International Conference on Safety Produce Informatization (IICSPI), (2019), 311–314. https://doi.org/10.1109/IICSPI48186.2019.9096014
    [5] S. Li, J. Zhang, J. Zhang, L. Sun, Y. Liu, Study on the secant segmentation algorithm of rubber tree, J. Phys. Conf. Ser., 1004 (2018), 012033. https://doi.org/10.1088/1742-6596/1004/1/012033 doi: 10.1088/1742-6596/1004/1/012033
    [6] P. Parvati, B. Rao, M. Das, Image segmentation using gray-scale morphology and marker-controlled watershed transformation, Discrete Dyn. Nat. Soc., 2008 (2008). https://doi.org/10.1155/2008/384346 doi: 10.1155/2008/384346
    [7] P. Sathya, R. Kalyani, V. Sakthivel, Color image segmentation using kapur, otsu and minimum cross entropy functions based on exchange market algorithm, Expert Syst. Appl., 172 (2021), 114636. https://doi.org/10.1016/j.eswa.2021.114636 doi: 10.1016/j.eswa.2021.114636
    [8] X. Wang, S. Wang, Y. Guo, K. Hu, W. Wang, Coal gangue image segmentation method based on edge detection theory of star algorithm, Int. J. Coal Prep. Util., 43 (2023), 119–134. https://doi.org/10.1080/19392699.2021.2024173 doi: 10.1080/19392699.2021.2024173
    [9] H. Yu, P. Sun, F. He, Z. Hu, A weighted region-based level set method for image segmentation with intensity inhomogeneity, PLoS One, 16 (2021), e0255948. https://doi.org/10.1371/journal.pone.0255948 doi: 10.1371/journal.pone.0255948
    [10] D. Wei, Z. Wang, L. Si, C. Tan, X. Lu, An image segmentation method based on a modified local-information weighted intuitionistic fuzzy c-means clustering and gold-panning algorithm, Eng. Appl. Artif. Intell., 101 (2021), 104209. https://doi.org/10.1016/j.engappai.2021.104209 doi: 10.1016/j.engappai.2021.104209
    [11] P. Ghamisi, M. Couceiro, J. Benediktsson, N. Ferreira, An efficient method for segmentation of image based on fractional calculus and natural selection, Expert Syst. Appl., 39 (2012), 12407–12417. https://doi.org/10.1016/j.eswa.2012.04.078 doi: 10.1016/j.eswa.2012.04.078
    [12] K. Kumar, K. Venkatalakshmi, K. Krishnan, Lung cancer detection using image segmentation by means of various evolutionary algorithms, Comput. Math. Methods Med., 2019 (2019), 1–16. https://doi.org/10.1155/2019/4909846 doi: 10.1155/2019/4909846
    [13] Z. Wakaf, H. Jalab, Defect detection based on extreme edge of defective region histogram, J. King Saud Univ. Comput. Inf. Sci., 30 (2018), 33–40. https://doi.org/10.1016/j.jksuci.2016.11.001 doi: 10.1016/j.jksuci.2016.11.001
    [14] L. Zhang, A. Li, X. Li, S. Xu, X. Yang, Remote sensing image segmentation based on an improved 2-d gradient histogram and mmad model, IEEE Geosci. Remote Sens. Lett., 12 (2015), 58–62. https://doi.org/10.1109/LGRS.2014.2326008 doi: 10.1109/LGRS.2014.2326008
    [15] Y. Xie, L. Ning, M. Wang, C. Li, Image enhancement based on histogram equalization, J. Phys. Conf. Ser., 1314 (2019), 012161. https://doi.org/10.1088/1742-6596/1314/1/012161 doi: 10.1088/1742-6596/1314/1/012161
    [16] M. Abd Elaziz, A. A. Ewees, D. Oliva, Hyper-heuristic method for multilevel thresholding image segmentation, Expert Syst. Appl., 146 (2020), 113201. https://doi.org/10.1016/j.eswa.2020.113201 doi: 10.1016/j.eswa.2020.113201
    [17] S. Aja-Fernández, A. H. Curiale, G. Vegas-Sánchez-Ferrero, A local fuzzy thresholding methodology for multiregion image segmentation, Knowl.-Based Syst., 83 (2015), 1–12. https://doi.org/10.1016/j.knosys.2015.02.029 doi: 10.1016/j.knosys.2015.02.029
    [18] N. Otsu, A threshold selection method from gray-level histograms, IEEE Trans. Syst. Man Cybern., 9 (1979), 62–66. https://doi.org/10.1109/TSMC.1979.4310076 doi: 10.1109/TSMC.1979.4310076
    [19] J. N. Kapur, P. K. Sahoo, A. K. Wong, A new method for gray-level picture thresholding using the entropy of the histogram, Comput. Vision, Graphics, Image Process., 29 (1985), 273–285. https://doi.org/10.1016/0734-189X(85)90125-2 doi: 10.1016/0734-189X(85)90125-2
    [20] S. Agrawal, R. Panda, S. Bhuyan, B. K. Panigrahi, Tsallis entropy based optimal multilevel thresholding using cuckoo search algorithm, Swarm Evol. Comput., 11 (2013), 16–30. https://doi.org/10.1016/j.swevo.2013.02.001 doi: 10.1016/j.swevo.2013.02.001
    [21] C. H. Li, C. Lee, Minimum cross entropy thresholding, Pattern Recognit., 26 (1993), 617–625. https://doi.org/10.1016/0031-3203(93)90115-D doi: 10.1016/0031-3203(93)90115-D
    [22] P. Y. Yin, Multilevel minimum cross entropy threshold selection based on particle swarm optimization, Appl. Math. Comput., 184 (2007), 503–513. https://doi.org/10.1016/j.amc.2006.06.057 doi: 10.1016/j.amc.2006.06.057
    [23] D. Wolpert, W. Macready, No free lunch theorems for optimization, IEEE Trans. Evol. Comput., 1 (1997), 67–82. https://doi.org/10.1109/4235.585893 doi: 10.1109/4235.585893
    [24] W. T. Pan, A new fruit fly optimization algorithm: Taking the financial distress model as an example, Knowl.-Based Syst., 26 (2012), 69–74. https://doi.org/10.1016/j.knosys.2011.07.001 doi: 10.1016/j.knosys.2011.07.001
    [25] A. A. Heidari, S. Mirjalili, H. Faris, I. Aljarah, M. Mafarja, H. Chen, Harris hawks optimization: Algorithm and applications, Future Gener. Comput. Syst., 97 (2019), 849–872. https://doi.org/10.1016/j.future.2019.02.028 doi: 10.1016/j.future.2019.02.028
    [26] S. Li, H. Chen, M. Wang, A. A. Heidari, S. Mirjalili, Slime mould algorithm: A new method for stochastic optimization, Future Gener. Comput. Syst., 111 (2020), 300–323. https://doi.org/10.1016/j.future.2020.03.055 doi: 10.1016/j.future.2020.03.055
    [27] Y. Feng, S. Deb, G. G. Wang, A. H. Alavi, Monarch butterfly optimization: A comprehensive review, Expert Syst. Appl., 168 (2021), 114418. https://doi.org/10.1016/j.eswa.2020.114418 doi: 10.1016/j.eswa.2020.114418
    [28] J. Tu, H. Chen, M. Wang, A. Gandomi, The colony predation algorithm, J. Bionic Eng., 18 (2021), 674–710. https://doi.org/10.1007/s42235-021-0050-y doi: 10.1007/s42235-021-0050-y
    [29] G. G. Wang, Moth search algorithm: a bio-inspired metaheuristic algorithm for global optimization problems, Memet. Comput., 10 (2018), 151–164. https://doi.org/10.1007/s12293-016-0212-3 doi: 10.1007/s12293-016-0212-3
    [30] Y. Yang, H. Chen, A. A. Heidari, A. H. Gandomi, Hunger games search: Visions, conception, implementation, deep analysis, perspectives, and towards performance shifts, Expert Syst. Appl., 177 (2021), 114864. https://doi.org/10.1016/j.eswa.2021.114864 doi: 10.1016/j.eswa.2021.114864
    [31] I. Ahmadianfar, A. A. Heidari, S. Noshadian, H. Chen, A. H. Gandomi, Info: An efficient optimization algorithm based on weighted mean of vectors, Expert Syst. Appl., 195 (2022), 116516. https://doi.org/10.1016/j.eswa.2022.116516 doi: 10.1016/j.eswa.2022.116516
    [32] I. Ahmadianfar, A. A. Heidari, A. H. Gandomi, X. Chu, H. Chen, Run beyond the metaphor: An efficient optimization algorithm based on runge kutta method, Expert Syst. Appl., 181 (2021), 115079. https://doi.org/10.1016/j.eswa.2021.115079 doi: 10.1016/j.eswa.2021.115079
    [33] M. A. Awadallah, M. A. Al-Betar, M. S. Braik, A. I. Hammouri, I. A. Doush, R. A. Zitar, An enhanced binary rat swarm optimizer based on local-best concepts of pso and collaborative crossover operators for feature selection, Comput. Biol. Med., 147 (2022), 105675. https://doi.org/10.1016/j.compbiomed.2022.105675 doi: 10.1016/j.compbiomed.2022.105675
    [34] S. Thawkar, S. Sharma, M. Khanna, L. kumar Singh, Breast cancer prediction using a hybrid method based on butterfly optimization algorithm and ant lion optimizer, Comput. Biol. Med., 139 (2021), 104968. https://doi.org/10.1016/j.compbiomed.2021.104968 doi: 10.1016/j.compbiomed.2021.104968
    [35] S. Chakraborty, A. K. Saha, S. Nama, S. Debnath, Covid-19 x-ray image segmentation by modified whale optimization algorithm with population reduction, Comput. Biol. Med., 139 (2021), 104984. https://doi.org/10.1016/j.compbiomed.2021.104984 doi: 10.1016/j.compbiomed.2021.104984
    [36] G. I. Sayed, M. M. Soliman, A. E. Hassanien, A novel melanoma prediction model for imbalanced data using optimized squeezenet by bald eagle search optimization, Comput. Biol. Med., 136 (2021), 104712. https://doi.org/10.1016/j.compbiomed.2021.104712 doi: 10.1016/j.compbiomed.2021.104712
    [37] M. A. Awadallah, A. I. Hammouri, M. A. Al-Betar, M. S. Braik, M. A. Elaziz, Binary horse herd optimization algorithm with crossover operators for feature selection, Comput. Biol. Med., 141 (2022), 105152. https://doi.org/10.1016/j.compbiomed.2021.105152 doi: 10.1016/j.compbiomed.2021.105152
    [38] J. K. Xue, B. Shen, A novel swarm intelligence optimization approach: sparrow search algorithm, Syst. Sci. Control Eng., 8 (2020), 22–34. https://doi.org/10.1080/21642583.2019.1708830 doi: 10.1080/21642583.2019.1708830
    [39] S. Mirjalili, S. Mirjalili, A. Lewis, Grey wolf optimizer, Adv. Eng. Software, 69 (2014), 46–61. https://doi.org/10.1016/j.advengsoft.2013.12.007 doi: 10.1016/j.advengsoft.2013.12.007
    [40] S. Mirjalili, A. Lewis, The whale optimization algorithm, Adv. Eng. Software, 95 (2016), 51–67. https://doi.org/10.1016/j.advengsoft.2016.01.008 doi: 10.1016/j.advengsoft.2016.01.008
    [41] C. Huang, X. Li, Y. Wen, An otsu image segmentation based on fruitfly optimization algorithm, Alexandria Eng. J., 60 (2021), 183–188. https://doi.org/10.1016/j.aej.2020.06.054 doi: 10.1016/j.aej.2020.06.054
    [42] A. Bhandari, A. Kumar, G. Singh, Modified artificial bee colony based computationally efficient multilevel thresholding for satellite image segmentation using Kapur's, otsu and tsallis functions, Expert Syst. Appl., 42 (2015), 1573–1601. https://doi.org/10.1016/j.eswa.2014.09.049 doi: 10.1016/j.eswa.2014.09.049
    [43] E. Houssein, D. Abdelkareem, M. Emam, M. Hameed, M. Younan, An efficient image segmentation method for skin cancer imaging using improved golden jackal optimization algorithm, Comput. Biol. Med., 149 (2022), 106075. https://doi.org/10.1016/j.compbiomed.2022.106075 doi: 10.1016/j.compbiomed.2022.106075
    [44] Z. Zhang, J. Yin, Bee foraging algorithm based multi-level thresholding for image segmentation, IEEE Access, 8 (2020), 16269–16280. https://doi.org/10.1109/ACCESS.2020.2966665 doi: 10.1109/ACCESS.2020.2966665
    [45] M. Abdel-Basset, R. Mohamed, N. AbdelAziz, M. Abouhawwash, Hwoa: A hybrid whale optimization algorithm with a novel local minima avoidance method for multi-level thresholding color image segmentation, Expert Syst. Appl., 190 (2021), 116145. https://doi.org/10.1016/j.eswa.2021.116145 doi: 10.1016/j.eswa.2021.116145
    [46] G. Kang, S. Gao, L. Yu, D. Zhang, Deep architecture for high-speed railway insulator surface defect detection: Denoising autoencoder with multitask learning, IEEE Trans. Instrum. Meas., 68 (2018), 2679–2690. https://doi.org/10.1109/TIM.2018.2868490 doi: 10.1109/TIM.2018.2868490
    [47] Y. Zhan, G. Zhang, An improved otsu algorithm using histogram accumulation moment for ore segmentation, Symmetry, 11 (2019), 431. https://doi.org/10.3390/sym11030431 doi: 10.3390/sym11030431
    [48] X. Xu, S. Xu, L. Jin, E. Song, Characteristic analysis of otsu threshold and its applications, Pattern Recognit. Lett., 32 (2011), 956–961. https://doi.org/10.1016/j.patrec.2011.01.021 doi: 10.1016/j.patrec.2011.01.021
    [49] S. Tripathi, K. Kumar, B. Singh, R. Singh, Image segmentation: A review, Int. J. Comput. Sci. Manage. Res., 1 (2012), 838–843.
    [50] P. Sathya, R. Kayalvizhi, Optimal multilevel thresholding using bacterial foraging algorithm, Expert Syst. Appl., 38 (2011), 15549–15564. https://doi.org/10.1016/j.eswa.2011.06.004 doi: 10.1016/j.eswa.2011.06.004
    [51] F. A.Hashim, A. Hussien, Snake optimizer: A novel meta-heuristic optimization algorithm, Knowl.-Based Syst., 242 (2022), 108320. https://doi.org/10.1016/j.knosys.2022.108320 doi: 10.1016/j.knosys.2022.108320
    [52] L. Qingge, R. Zheng, X. Zhao, S. Wei, P. Yang, An improved otsu threshold segmentation algorithm, Int. J. Comput. Sci. Eng., 22 (2020), 146–153. https://doi.org/10.1504/IJCSE.2020.107266 doi: 10.1504/IJCSE.2020.107266
    [53] M. H. Horng, Multilevel thresholding selection based on the artificial bee colony algorithm for image segmentation, Expert Syst. Appl., 38 (2011), 13785–13791. https://doi.org/10.1016/j.eswa.2011.04.180 doi: 10.1016/j.eswa.2011.04.180
    [54] S. Sarkar, S. Das, S. Chaudhuri, A multilevel color image thresholding scheme based on minimum cross entropy and differential evolution, Pattern Recognit. Lett., 54 (2015), 27–35. https://doi.org/10.1016/j.patrec.2014.11.009 doi: 10.1016/j.patrec.2014.11.009
    [55] D. Oliva, V. Osuna-Enciso, E. Cuevas, G. Pajares, M. Cisneros, D. Zaldivar, Improving segmentation velocity using an evolutionary method, Expert Syst. Appl., 42 (2015), 5874–5886. https://doi.org/10.1016/j.eswa.2015.03.028 doi: 10.1016/j.eswa.2015.03.028
  • This article has been cited by:

    1. Lusha Shi, Jianghong Hu, Zhen Jin, Dynamics analysis of strangles with asymptomatic infected horses and long-term subclinical carriers, 2023, 20, 1551-0018, 18386, 10.3934/mbe.2023817
    2. Guijie Lan, Sanling Yuan, Baojun Song, Threshold behavior and exponential ergodicity of an sir epidemic model: the impact of random jamming and hospital capacity, 2024, 88, 0303-6812, 10.1007/s00285-023-02024-1
    3. Indunil M. Hewage, Kevin E. M. Church, Elissa J. Schwartz, Investigating the impact of vaccine hesitancy on an emerging infectious disease: a mathematical and numerical analysis, 2024, 18, 1751-3758, 10.1080/17513758.2023.2298988
    4. Hui Cao, Baojun Song, Yicang Zhou, Treatment strategies for the latent tuberculosis infections, 2023, 86, 0303-6812, 10.1007/s00285-023-01927-3
    5. Marcelo A. Pires, Cesar I.N. Sampaio Filho, Hans J. Herrmann, José S. Andrade, Tricritical behavior in epidemic dynamics with vaccination, 2023, 174, 09600779, 113761, 10.1016/j.chaos.2023.113761
    6. Qiuyun Li, Fengna Wang, An Epidemiological Model for Tuberculosis Considering Environmental Transmission and Reinfection, 2023, 11, 2227-7390, 2423, 10.3390/math11112423
    7. Yousef Alnafisah, Moustafa El-Shahed, Optimal control of red palm weevil model incorporating sterile insect technique, mechanical injection, and pheromone traps, 2024, 93, 11100168, 382, 10.1016/j.aej.2024.02.059
    8. Shaoli Wang, Tengfei Wang, Ya-Nen Qi, Fei Xu, Backward bifurcation, basic reinfection number and robustness of an SEIRE epidemic model with reinfection, 2023, 16, 1793-5245, 10.1142/S1793524522501327
    9. Lucero Rodriguez Rodriguez, Marisabel Rodriguez Messan, Komi Messan, Yun Kang, Mathematical modeling of natural resource and human interaction: Applications to the harvesting of Pacific Yew for cancer treatment, 2025, 0, 2155-3289, 0, 10.3934/naco.2025003
  • Reader Comments
  • © 2023 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(2252) PDF downloads(137) Cited by(6)

Figures and Tables

Figures(18)  /  Tables(4)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog