--- title: Two-Dimensional Smoothing Techniques url: https://www.emergentmind.com/topics/two-dimensional-smoothing-tds type: topic --- # Two-Dimensional Smoothing Techniques Searching arXiv for recent and representative papers on “Two-Dimensional Smoothing” and closely related usages. Two-Dimensional Smoothing (TDS) denotes a class of procedures that modify a two-dimensional object under explicit smoothness criteria while retaining application-dependent structure. In the literature, the object being smoothed may be a two-dimensional sequence, an image, a planar triangular mesh, a reduced potential-energy surface, a regression surface, or a dynamical state measured in a stronger function space. The operational meaning of “smoothing” therefore varies: it can mean suppressing fluctuation by penalizing second differences, selecting directionally compatible local averages, relocating mesh vertices only when local quality improves, constructing a continuous low-energy surface through an extra dimension, or proving that a two-dimensional flow map becomes regular from \(H\) into \(H^2\) after positive pullback time [2507.15301], [1608.01993], [2111.06513], [2508.18745]. ## 1. Scope and principal meanings Across the cited work, TDS is not a single algorithm but a family of two-dimensional regularization ideas. In one line of work, the data are arranged as \(Z=(z_{i,j})\in\mathbb{R}^{m\times n}\) and the trend \(G=(g_{i,j})\) is defined as the minimizer of a global loss combining fidelity and row/column second-difference penalties [2507.15301]. In SAR despeckling, directional smoothing computes several directional local averages in a moving \(2\text{D}\) window and chooses the direction whose average is closest to the center pixel, thereby smoothing along local structure rather than across edges [1608.01993]. In mesh optimization, two-dimensional Smart Laplacian smoothing relocates vertices of planar triangular meshes while preserving connectivity and accepts a move only if the local mesh quality improves [1502.00355]. In nuclear-fission calculations, two-dimensional smoothing refers to replacing a discontinuous \(2\text{D}\) potential-energy surface by a continuous low-energy surface \(z(x,y)\) inside a local \(3\text{D}\) search region [2111.06513]. In statistics, it includes tensor-product P-splines, thin-plate splines, smoothing thin plate splines with \(\mathcal H\)-matrix compression, and triangulation splines on polygonal domains [2211.16218], [2404.01902], [2606.12296]. In PDE and multigrid analysis, smoothing denotes either a regularity lift, such as \((H,H^2)\)-smoothing for stochastic \(2\text{D}\) Navier–Stokes, or damping of high-frequency error components by a multigrid relaxation scheme [2508.18745], [2306.17040]. | Setting | Smoothed object | Governing idea | |---|---|---| | 2D sequences and images | \(Z\), image intensity field | second-difference penalization or directional local averaging | | Planar triangular meshes | vertex coordinates | move accepted only if local quality improves | | Potential-energy surfaces | \(z(x,y)\) in a 3D search box | lowest-energy continuous surface under continuity constraints | | Statistical surface estimation | \(f(x_1,x_2)\) | spline basis plus roughness penalty | | PDE and multigrid | solution differences or error modes | regularity gain or high-frequency damping | This range suggests that TDS is best understood as a methodological category unified by two-dimensional structure and explicit smoothness control, rather than by a unique algebraic template. ## 2. Global and local formulations on two-dimensional sequences and images The most explicit algorithm named TDS in the cited corpus is the global second-difference formulation for a general two-dimensional sequence \(Z=(z_{i,j})\in\mathbb{R}^{m\times n}\), \(m,n\ge 3\). It posits a decomposition \[ Z=G+C, \] where \(G\) is the trend sequence and \(C\) is the fluctuation sequence, and defines \(G\) as the minimizer of \[ \begin{aligned} M =&\sum_{i=1}^{m}\sum_{j=1}^{n}\left(z_{i,j}-g_{i,j}\right)^2 \\ &+\lambda \left\{ \sum_{i=1}^{m}\sum_{j=3}^{n}\left[\Delta^2 g_{i,j}\right]^2 +\sum_{j=1}^{n}\sum_{i=3}^{m}\left[\nabla^2 g_{i,j}\right]^2 \right\}, \end{aligned} \] with first differences \(\Delta g_{i,j}=g_{i,j}-g_{i,j-1}\), \(\nabla g_{i,j}=g_{i,j}-g_{i-1,j}\), second differences \[ \Delta^2 g_{i,j}=g_{i,j}-2g_{i,j-1}+g_{i,j-2},\qquad \nabla^2 g_{i,j}=g_{i,j}-2g_{i-1,j}+g_{i-2,j}, \] and global smoothing parameter \(\lambda>0\) [2507.15301]. The normal equation is \[ G+\lambda(GT+HG)=Z, \] where \(T\) and \(H\) are sparse real symmetric positive semidefinite penalty matrices. The same work proves existence and uniqueness for fixed \(\lambda\), gives the vectorized solution \[ \mathrm{vec}(G)= \left(I_n\otimes I_m+\lambda T\otimes I_m+\lambda I_n\otimes H\right)^{-1}\mathrm{vec}(Z), \] and defines the fluctuation sequence as \[ C=Z-G=\lambda(GT+HG) \] [2507.15301]. It further introduces TDS-I, TDS-II, and TDS-III, which replace the single \(\lambda\) by directional global parameters \((\gamma,\delta)\) or row-/column-specific diagonal parameter matrices \((\Gamma,\Delta)\). A distinct image-oriented formulation appears in directional smoothing for SAR despeckling. There the observed image obeys the multiplicative model \[ v(r,c)=u(r,c)s(r,c), \] and smoothing is performed in a local \(2\text{D}\) window by computing directional averages \[ d(1)=\frac{v(r,c-1)+v(r,c+1)}{2},\quad d(2)=\frac{v(r-1,c)+v(r+1,c)}{2}, \] \[ d(3)=\frac{v(r-1,c-1)+v(r+1,c+1)}{2},\quad d(4)=\frac{v(r+1,c-1)+v(r-1,c+1)}{2}, \] then selecting \[ n^*=\arg\min_{n\in\{1,2,3,4\}} |d(n)-v(r,c)|,\qquad \hat v(r,c)=d(n^*) \] [1608.01993]. This is a nonlinear adaptive directional smoother rather than a global optimizer. The same paper applies the method homomorphically in the log domain and evaluates it using NV, MSD, ENL, DR, and Pratt figure of merit [1608.01993]. These two strands illustrate a basic division inside image and array smoothing. One class uses a global regularized objective with a single solution \(G\); the other uses local directional candidates and a nonlinear selection rule. This suggests that, even within classical data smoothing, TDS spans both variational and local-compatibility paradigms. ## 3. Geometry-aware smoothing of meshes and physical surfaces In planar mesh optimization, two-dimensional Smart Laplacian smoothing operates on a triangular mesh without changing connectivity. Ordinary Laplacian smoothing moves a vertex to the geometric center of its neighboring vertices, but it may degrade local quality and even create invalid or inverted elements. Smart Laplacian smoothing adds a quality-based acceptance test: evaluate the quality of the incident triangles before relocation, compute the Laplacian candidate position, re-evaluate the incident-triangle qualities at that position, and accept the move only if the local mesh quality improves; the explicit example criterion is that the minimum quality among all incident triangles increases [1502.00355]. The GPU implementation described there organizes the procedure into initiating, neighbor finding, constraint determination, and iterative smoothing, compares AoS and SoA layouts, compares Form A \[ \overline {x_i^{q+1} } =\frac{1}{N}\sum\limits_{j=1}^N {x_j^q } \] with Form B \[ \overline {x_i^{q+1} } =\frac{1}{N}\left( {\sum\limits_{j=1}^{N_q } {x_j^q } +\sum\limits_{k=1}^{N_{q+1} } {x_k^{q+1} } } \right), \] and reports that AoS is always faster than SoA, Form B is faster than Form A, and the implementation reaches speedups up to \(44.76\times\) on a GeForce GT640 [1502.00355]. A different geometric meaning of TDS appears in nuclear-fission potential-energy surfaces. There, the problem is a discontinuous reduced \(2\text{D}\) PES in coordinates \((x,y)\), caused by an omitted coordinate \(z\). The proposed Frontier DPM seeks a continuous surface \(z(x,y)\) inside a local \(3\text{D}\) search space, minimizing total energy while enforcing a maximum gradient constraint \(\Delta_{\max}\) and an overlap threshold \(\mathcal N_{\min}\) between adjacent HFB states [2111.06513]. The method represents partial surfaces by their frontiers, expands them one planar site at a time, and prunes states with identical frontier configurations by keeping only the lowest-energy representative. The brute-force search space scales as \(L^{MN}\), whereas Frontier DPM reduces dependence to the frontier width, with an estimated cost on the order of \(N M L^{M+1}\) or the analogous expression with \(M\) and \(N\) interchanged [2111.06513]. In the \(^{218}\mathrm{Ra}\) example, it removes a diagonal discontinuity in \(Q_{40}\), eliminates zero-overlap adjacencies, and reduces density distances [2111.06513]. These uses show that geometric TDS is not primarily about denoising. In mesh processing it is a quality-monotonic relocation rule on a fixed connectivity graph; in nuclear PES repair it is a constrained search for a physically continuous low-energy surface embedded in a higher-dimensional space. ## 4. Statistical surface smoothing: tensor products, thin plates, and triangulations A central statistical formulation of TDS is the tensor-product spline surface \[ f(x_1,x_2)\approx \sum_{r=1}^{d_1}\sum_{s=1}^{d_2} b_{rs}\,B_r^{(1)}(x_1)\,B_s^{(2)}(x_2), \] with anisotropic penalty \[ K(\tau_1^2,\tau_2^2) = \frac{1}{\tau_1^2}\big(\widetilde K_1\otimes I_{d_2}\big) + \frac{1}{\tau_2^2}\big(I_{d_1}\otimes \widetilde K_2\big) \] [2211.16218]. In the Bayesian formulation, the spline coefficients have partially improper Gaussian prior \[ p(b\mid \tau_1^2,\tau_2^2)\propto Det(K)^{1/2}\exp\!\left(-\frac12 b^\top K b\right), \] and the technical bottleneck is the pseudo-determinant in the smoothing-parameter posterior. The paper resolves this by deriving closed-form expressions for \(\log Det(K(\tau^2))\) and its derivatives using Kronecker-structured eigendecompositions, enabling adaptive Metropolis-Hastings updates for \((\tau_1^2,\tau_2^2)\) in arbitrary dimension and making the \(p=2\) case computationally straightforward [2211.16218]. Thin-plate smoothing supplies a second major line. Robust thin-plate splines replace quadratic data fit by a convex M-estimation loss, \[ L_n(f)=\frac{1}{n}\sum_{i=1}^n \rho(Y_i-f(\mathbf{x}_i)) + \lambda I_m^2(f), \] where, in the \(d=2,m=2\) case, \[ I_2^2(f) = \iint \left[ \left(\frac{\partial^2 f}{\partial x_1^2}\right)^2 +2\left(\frac{\partial^2 f}{\partial x_1\partial x_2}\right)^2 +\left(\frac{\partial^2 f}{\partial x_2^2}\right)^2 \right]dx_1dx_2 \] [2207.13931]. The representer form remains \[ f(\mathbf{x}) = \sum_{i=1}^n \gamma_i \eta_{m,d}(\|\mathbf{x}-\mathbf{x}_i\|) + \sum_{j=1}^M \delta_j \phi_j(\mathbf{x}), \] which in the classical two-dimensional thin-plate case corresponds to an affine polynomial plus the \(r^2\log r\) kernel. The same paper proves optimal convergence rates under mild assumptions and emphasizes invariance of the thin-plate penalty under rigid transformations [2207.13931]. Classical smoothing thin plate splines become computationally demanding because the kernel matrix is dense. An efficient realization compresses the dense TPS matrix by an \(\mathcal H\)-matrix and solves the resulting Schur-complement SPD system by conjugate gradients [2404.01902]. In that work the smoothing objective is \[ S(g) = \sum^{n}_{i = 1} |y_{i} - g(\boldsymbol{x}_{i})|^{2} + \lambda J(g), \] with \(J(g)\) the two-dimensional bending-energy penalty, and the dense kernel matrix \(\boldsymbol E\) is approximated blockwise by low-rank admissible far-field blocks using Adaptive Cross Approximation [2404.01902]. The reported interpolation RMSE on an \(80^2\) training grid and \(40^2\) prediction grid is \(0.01\) for the exact method, the Schur-complement CG method, and the \(\mathcal H\)-matrix method, while the latter exhibits practically linear growth in the reported timing plots and strong sensitivity to the ACA tolerance \(\varepsilon\) [2404.01902]. For irregular polygonal domains, Bayesian triangulation splines replace tensor products and global radial kernels by spline spaces on constrained Delaunay triangulations. The domain \(\Omega\subset\mathbb R^2\) is triangulated, the spline space is \[ \mathcal{S}_d^r(\Delta):=\{g\in C^r(\Omega): g|_T\in \mathbb{P}_d(T),\ T\in\Delta\}, \] and the roughness penalty is the thin-plate functional \[ \mathcal{E}(f)=\int_\Omega \left\{(\nabla_{11}^2 f)^2+2(\nabla_{12}^2 f)^2+(\nabla_{22}^2 f)^2\right\}\,ds_1ds_2 \] [2606.12296]. The prior is placed jointly on triangulation complexity and coefficient vectors, which allows adaptation to unknown global smoothness and to local spatial inhomogeneity; the paper proves optimal posterior contraction under a global Sobolev assumption and an oracle-type spatial adaptation guarantee over shape-regular triangulations [2606.12296]. Taken together, these statistical formulations show that TDS can be global and Euclidean, robust to contamination, computationally compressed, or boundary-conforming and locally adaptive. The choice among them is governed less by the word “smoothing” than by domain geometry, robustness requirements, and computational regime. ## 5. Smoothing as regularity gain and multigrid damping in two dimensions In PDE and dynamical-systems analysis, smoothing has a different meaning from denoising or surface fitting. For stochastic Navier–Stokes on the two-dimensional torus \(\mathbb T^2\), the relevant result is an \((H,H^2)\)-smoothing theorem: for tempered initial sets in the energy space \(H\), there exist random variables \(T_{\mathfrak D}(\omega)\) and \(L_{\mathfrak D}(\omega)\) such that \[ \big\|v_1(T_{\mathfrak D}(\omega),\theta_{-T_{\mathfrak D}(\omega)}\omega,v_{1,0}) -v_2(T_{\mathfrak D}(\omega),\theta_{-T_{\mathfrak D}(\omega)}\omega,v_{2,0}) \big \|^2_{H^2} \leq L_{\mathfrak D} (\omega)\|v_{1,0}-v_{2,0}\|^2_H. \] This is established under \(f\in H\), \(h\in H^3\), and \[ \|\nabla h\|_{L^\infty}<\sqrt{\pi}\,\nu\,\lambda_1, \] and is then used to promote the random attractor from \(H\) to a tempered \((H,H^2)\)-random attractor of finite fractal dimension in \(H^2\) [2508.18745]. Here smoothing means that the cocycle becomes locally Lipschitz from a weaker phase space into a stronger one after positive pullback time. In multigrid analysis for \(2\text{D}\) Stokes flow, smoothing refers to damping high-frequency error. The stabilized non-staggered discretization \[ L_h= \begin{pmatrix} -\Delta_h & 0 & \partial_x^h\\ 0 & -\Delta_h & \partial_y^h\\ \partial_x^h & \partial_y^h & -ch^2\Delta_h \end{pmatrix} \] is transformed by a distributive operator so that the smoothing problem reduces to the scalar blocks \(-\Delta_h\) and \(ch^2\Delta_h^2-\Delta_h\) [2306.17040]. The LFA smoothing factor is \[ \mu(n,L_h)=\max\{\mu(n,-\Delta_h),\mu(n,ch^2\Delta_h^2-\Delta_h)\}, \] the Poisson block yields \[ \omega_{\mathrm{opt}}=\frac{16}{17},\qquad P_{\mathrm{opt}}=\frac{1}{17}, \] and the stabilized pressure block yields a \(c\)-dependent optimal factor with an effective convergence zone approximately \[ 0