A class of thin-film equations with space-time dependent gradient nonlinearity and its application to image sharpening
Yuhang Li a {}^{\ \textrm{a}}, Zhichang Guo \cormark a , [ 1 ] {}^{\ \textrm{a},}\cormark[1], Fanghui Song a {}^{\ \textrm{a}}, Boying Wu a {}^{\ \textrm{a}}, Bin Guo b {}^{\ \textrm{b}} · 2025
Read A class of thin-film equations with space-time dependent gradient nonlinearity and its application to image sharpening by Yuhang Li a {}^{\ \textrm{a}}, Zhichang Guo \cormark a , [ 1 ] {}^{\ \textrm{a},}\cormark[1], Fanghui Song a {}^{\ \textrm{a}}, Boying Wu a {}^{\ \textrm{a}}, Bin Guo b {}^{\ \textrm{b}} (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-19211.txt.
Introduction
Yuhang Li and co-authors apply a film equation to image sharpening in a 2025 paper. Thin-film equations track how a surface layer flows and levels. With a slope term that shifts over space and time, they gain new uses beyond physics.
The team sort when solutions last for all time and when they blow up in finite time. A reshaped well method handles the moving sets that variable terms create. Decay rates for lasting solutions and lifespan bounds for blow-ups are also set. Image tests then show the equations sharpen pictures well. The work joins hard analysis to a plain, useful task.
Catalog record
| CREATOR | Yuhang Li a {}^{\ \textrm{a}}, Zhichang Guo \cormark a , [ 1 ] {}^{\ \textrm{a},}\cormark[1], Fanghui Song a {}^{\ \textrm{a}}, Boying Wu a {}^{\ \textrm{a}}, Bin Guo b {}^{\ \textrm{b}} |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 102 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-19211
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A class of thin-film equations with space-time dependent gradient nonlinearity and its application to image sharpening Yuhang Li a {}^{\ \textrm{a}}, Zhichang Guo \cormark a , [ 1 ] {}^{\ \textrm{a},}\cormark[1], Fanghui Song a {}^{\ \textrm{a}}, Boying Wu a {}^{\ \textrm{a}}, Bin Guo b {}^{\ \textrm{b}} Abstract We introduce a class of thin-film equations with space-time dependent gradient nonlinearity and apply them to image sharpening. By modifying the potential well method, we overcome the challenges arising from variable exponents and the “moving" Nehari manifold, thereby providing a classification of global existence and finite time blow-up of solutions under different initial conditions. Decay rates for global solutions and lifespan estimates for blow-up solutions are also established. Numerical experiments demonstrate the effectiveness of this class of equations as image sharpening models. Journal: arXiv 1 Introduction Let \Omega\subset\mathbb{R}^{N} ( N\geq 2 ) be a bounded domain. For \alpha\in\mathbb{R} , we consider the following initial-boundary value problem: \left\{\begin{array}[]{ll}u_{t}+\mathcal{L}_{\alpha}u=\operatorname{div}\left(F(t,x,\nabla u)\right),&x\in\Omega,\ t>0,\\ u(0,x)=u_{0}(x),&x\in\Omega,\\ u=0,&x\in\mathbb{R}^{N}\backslash\Omega.\end{array}\right. (1.1) Here, \mathcal{L}_{\alpha}=(-\Delta)^{2}+\alpha(-\Delta)^{2s} is a mixed local-nonlocal biharmonic operator, s\in(0,1) , F(t,x,\xi)=\xi-k(t)|\xi|^{p(x)-2}\xi , p:\mathbb{R}^{N}\to(2,+\infty) . This type of problem appears in the modeling of epitaxial thin film growth. On a square domain, letting u(t,x) denote the height of the thin film, the continuum model for epitaxial thin film growth is given by [ 81 ] : u_{t}=-\operatorname{div}(\mathbf{j})+g+\eta, (1.2) where \mathbf{j}=\mathbf{j}(t,x) , g=g(t,x) , \eta=\eta(t,x) denotes all process of moving atoms along the surface, the deposition flux and Gaussian noise, respectively. One of the most widespread treatments is to neglect the noise term and assume [ 81 , 52 ] \mathbf{j}=A_{1}\nabla u+A_{2}\nabla(\Delta u)+A_{3}|\nabla u|^{p-2}\nabla u, so that ( 1.2 ) reduces to u_{t}+A_{1}\Delta u+A_{2}\Delta^{2}u+A_{3}\operatorname{div}\left(|\nabla u|^{p-2}\nabla u\right)=g. (1.3) In this case, the term A_{1}\Delta u corresponds to diffusion driven by evaporation-condensation, A_{2}\Delta^{2}u represents capillarity-driven surface diffusion, and A_{3}\operatorname{div}\left(|\nabla u|^{p-2}\nabla u\right) accounts for the upward hopping of atoms [ 51 , 15 ] . The intriguing mathematical properties of ( 1.3 ) have attracted extensive research interest over the past few decades. King et al. [ 37 ] studied the global existence and asymptotic behavior of solutions to ( 1.3 ) in the case where A_{1}>0,A_{2}>0,A_{3}<0 and g=0 . Later, Kohn et al. [ 38 ] analyzed the coarsening behavior of solutions under the same setting. Sandjo et al. [ 60 ] investigated the local well-posedness in L^{p} space of ( 1.3 ) when A_{1}=0,A_{2}>0,A_{3}\neq 0 ; afterward, Li et al. [ 44 ] established well-posedness results in local BMO space for p=4 and further studied the stability for global solutions for small initial data in VMO space. Related problems include variants of ( 1.3 ) involving nonlinear source terms [ 43 , 34 , 41 ] , time-dependent coefficients [ 56 , 57 , 84 ] , or Hessians [ 23 , 47 ] . In recent years, particular attention has been devoted to the case where A_{2}>0 and A_{3}>0 . Ishige [ 35 ] et al. studied the local existence and sufficient conditions for blow-up of solutions to the Cauchy problem for u_{t}+(-\Delta)^{2}u=-\operatorname{div}\left(|\nabla u|^{p-2}\nabla u\right), (1.4) on \mathbb{R}^{N} . Feng et al. [ 24 ] investigated the global existence and blow-up for ( 1.4 ) with and without advection on the torus \mathbb{T}^{2} . Miyake et al. [ 50 ] first established the asymptotic behavior of global solutions with subcritical initial energy under Neumann boundary conditions, and Zhao et al. [ 83 ] subsequently completed the classification of global existence and blow-up for this problem. The problem ( 1.1 ) studied in this paper corresponds to the case A_{1}<0 , A_{2}>0 and A_{3}>0 . Specifically, by setting \alpha=0 and taking F(t,x,\xi)=-|\xi|^{p-2}\xi in ( 1.1 ), we recover the form of ( 1.4 ). We claim that ( 1.1 ) can serve as an image sharpening model. Image sharpening refers to the enhancement of edges in an image to make it clearer and more visually appealing. Unlike image restoration, which emphasizes fidelity to a clean reference image, sharpening focuses more on improving the overall visual clarity [ 30 ] . It is commonly used in scenarios where images are intended for public display and aesthetic presentation. The simplest sharpening PDE is the linear backward diffusion equation, which is ill-posed and extremely sensitive to noise in the image, and thus requiring some form of stabilization or regularization techniques in practical computations [ 54 , 69 , 73 ] . Aside from linear backward diffusion, other well-known sharpening PDEs include nonlinear backward diffusion equations [ 58 , 3 , 62 ] , anisotropic forward-backward diffusion equations [ 28 , 29 , 7 ] , and a class of nonlinear hyperbolic equations known as shock filters [ 53 ] . Problem ( 1.1 ) involves a mixed local-nonlocal biharmonic operator and a space-time dependent nonlinear gradient term. Early works [ 79 , 46 , 4 , 33 , 42 ] employed fourth-order PDEs for image restoration, demonstrating their effectiveness in avoiding staircase effects and preserving ramps (affine regions). A recent study [ 66 ] applied a thin film-type equation to image segmentation and reported favorable results. As nonlocal operators have been found to be effective in texture preservation [ 27 , 9 , 68 , 78 , 71 ] , mixed local and nonlocal operators have recently attracted increasing attention in image processing [ 11 , 16 , 64 , 65 ] . Moreover, spatially or space-time dependent differential operators are commonly used in image processing models as they can more effectively exploit the intrinsic features of the image and improve the self-adaptability of models [ 13 , 32 , 70 , 63 , 76 ] . Inspired by the above works and ideas, we propose ( 1.1 ) as an image sharpening model, which can be analyzed as follows: if we disregard the term \mathcal{L}_{\alpha}u on the left-hand side and focus on the right-hand side term \operatorname{div}\left((1-k(t)|\nabla u|^{p(x)-2})\nabla u\right) , one can identify a space-time dependent threshold function \mathfrak{T}(t,x)=(k(t))^{\frac{1}{2-p(x)}} , such that: when |\nabla u|<\mathfrak{T}(t,x) , the diffusion coefficient 1-k(t)|\nabla u|^{p(x)-2} is positive, exhibiting forward diffusion behavior; when |\nabla u|>\mathfrak{T}(t,x) , the coefficient becomes negative, leading to backward diffusion. By appropriately designing the functions k(t) and p(x) , one can control the evolution process and ensure that backward diffusion occurs primarily near image edges, thereby achieving edge enhancement. The term \mathcal{L}_{\alpha}u , on the other hand, acts as a viscosity-type term that helps prevent the model from becoming ill-posed and reduces the influence of noise during the sharpening process. By tuning the parameter \alpha , one can achieve edge enhancement without amplifying noise in smooth regions, and maintaining fine details such as ramps and textures. The main contribution of this paper is the theoretical analysis of problem ( 1.1 ), especially the classification of global existence and finite time blow-up, which can be regarded as an extension and continuation of previous works [ 50 , 83 ] on ( 1.4 ). The analyses in [ 50 ] and [ 83 ] are mainly based on the potential well method, originally introduced by Sattinger and Payne [ 61 , 55 ] and modified by Liu et al. [ 45 ] , Gazzola et al. [ 26 ] and Xu et al. [ 74 ] . Besides