Search

Surface reconstruction for scattered data is an ill-posed problem and most computational algorithms become overly expensive as the number of sample points increases. This article studies an effecti...

This article is concerned with an effective PDE-based image zooming method which is an improvement of the so-called curvature interpolation method (CIM) previously developed in [H. Kim, Y. Cha, and S. Kim, IEEE Trans. Image Process., 20 (2011), pp. 1895--1903]. The new algorithm incorporates an iterative application of the CIM in order to be an interpolator and to construct a reliable image surface utilizing a generalized curvature source term estimated from the low resolution image. It also employs an effective 9-point scheme for the curvature evaluation which in turn makes the constructed image surface involve no observable ringing artifacts. It has been experimentally verified that the new method results in a high resolution image of clear and sharp edges and cloudless textures, showing qualities of superresolution. Its convergence is analyzed; various results are given to show effectiveness and reliability of the new method. The proposed method has proved not only superior to state-of-the-art methods ...

This article is concerned with effective numerical techniques for partial differential equation (PDE)-based image restoration. Numerical realizations of most PDE-based denoising models show a common drawback: loss of fine structures. In order to overcome the drawback, the article introduces a new time-stepping procedure, called the method of nonflat time evolution (MONTE), in which the timestep size is determined based on local image characteristics such as the curvature or the diffusion magnitude. The MONTE provides PDE-based restoration models with an effective mechanism for the equalization of the net diffusion over a wide range of image frequency components. It can be easily applied to diverse evolutionary PDE-based restoration models and their spatial and temporal discretizations. It has been numerically verified that the MONTE results in a significant reduction in numerical dissipation and preserves fine structures such as edges and textures satisfactorily, while it removes the noise with an improved efficiency. Various numerical results are shown to confirm the claim.

Image denoising processes often lead to significant loss of fine structures such as edges and textures. This paper studies various innovative mathematical and numerical methods applicable for conventional PDE-based denoising models. The method of diffusion modulation is considered to effectively minimize regions of undesired excessive dissipation. Then we introduce a novel numerical technique for residual-driven constraint parameterization, in order for the resulting algorithm to produce clear images whose corresponding residual is as free of image textures as possible. A linearized Crank-Nicolson alternating direction implicit time-stepping procedure is adopted to simulate the resulting model efficiently. Various examples are presented to show efficiency and reliability of the suggested methods in image denoising.

This article presents results of a parameter study for a new denoising model, using parallel computing and advanced dynamic load balancing techniques for performance improvement of implementations. A denoising model is suggested hybridizing total variation and Laplacian mean-curvature; the fourth-order model and its numerical procedure introduce a number of parameters. As a preliminary step in the model development, a parameter study has been undertaken in order to discover solitary and interactive effects of the parameters on model accuracy. Such a parameter study is necessarily time-consuming due to the huge number of combinations of the parameter values to be tested. In addition, the study has to be performed on various images, thereby increasing the overall investigation time. The performance of this first parallel implementation of a new hybrid model for image denoising is evaluated when the application is running on heterogeneous environments. The hybrid model is simulated on a general-purpose Linux cluster for which the parallel efficiency exceeds 96%.

This article introduces a new fourth-order implicit time-stepping scheme for the numerical solution of the acoustic wave equation, as a variant of the conventional modified equation method. For an efficient simulation, the scheme incorporates a locally one-dimensional (LOD) procedure having the splitting error of O(@Dt^4). Its stability and accuracy are compared with those of the standard explicit fourth-order scheme. It has been observed from various experiments for 2D problems that (a) the computational cost of the implicit LOD algorithm is only about 40% higher than that of the explicit method, for the problems of the same size, (b) the implicit LOD method produces less dispersive solutions in heterogeneous media, and (c) its numerical stability and accuracy match well those of the explicit method.

The article is concerned with edge-forming methods to be applied as a post-process for image zooming. Image zooming via standard interpolation methods often produces the so-called checkerboard effect, in particular, when the magnification factor is large. In order to remove the artifact and to form reliable edges, a nonlinear semi-discrete model and its numerical algorithm are suggested along with anisotropic edge-forming numerical schemes. The algorithm is analyzed for stability and choices of parameters. For image zooming by integer factors, a few iterations of the algorithm can form clear and sharp edges for gray-scale images. Various examples are presented to show effectiveness and efficiency of the newly-suggested edge-forming strategy.

The paper is concerned with PDE-based image restoration. A new model is introduced by hybridizing a nonconvex variant of the total variation minimization (TVM) and the motion by mean curvature (MMC) in order to deal with the mixture of the impulse and Gaussian noises reliably. We suggest the essentially nondissipative (ENoD) difference schemes for the MMC component to eliminate the impulse noise with a minimum (ideally no) introduction of dissipation. The MMC-TVM hybrid model and the ENoD schemes are applied for both gray-scale and color images. For color image denoising, we consider the chromaticity-brightness decomposition with the chromaticity formulated in the angle domain. An incomplete Crank-Nicolson alternating direction implicit time-stepping procedure is adopted to solve those differential equations efficiently. Numerical experiments have shown that the new hybrid model and the numerical schemes can remove the mixture of the impulse and Gaussian noises, efficiently and reliably, preserving edges quite satisfactorily.

Image denoising is still a challenging problem, particularly when the noise is made combining Gaussian noise and random-valued impulses. This article is concerned with diffusion-based denoising methods which can suppress such complicated noises effectively, preserving fine structures. We introduce a novel impulse-mowing anisotropic diffusion (IMAD) filter to cut out impulses and local maxima/minima without affecting surrounding pixel values. It has been numerically verified that the suggested mean filter carries out both mowing impulses and restoring fine structures satisfactorily. It outperforms nonlinear median filters, measured in PSNR and visual inspection.

The article is concerned with the development and comparison of three different algorithms for the computation of first‐arrival traveltimes: the fast marching method (FMM), the group marching method (GMM), and a second‐order finite‐difference eikonal solver. GMM is introduced as a variant of FMM. It proceeds the solution by advancing a selected group of grid points at a time, rather than sorting the solution in the narrow band to march forward a single grid point. The second‐order eikonal solver studied in the article is an expanding‐box, essentially nonoscillatory scheme for which the stability is enforced by the introduction of a down ‘n’ out marching and a post‐sweeping iteration.Techniques such as the maximum angle condition, the average normal velocity, and cache‐based implementation are introduced for the algorithms to improve the numerical accuracy and efficiency. The algorithms are implemented for solving the eikonal equation in 3‐D isotropic media, and their performances are compared. GMM is numerically verified to be faster than FMM. However, the second‐order algorithm turns out to be superior to these first‐order level‐set methods in both accuracy and efficiency; the incorporation of average normal velocity improves accuracy dramatically for the second‐order scheme.

We consider a second‐order finite difference scheme to solve the eikonal equation. Upwind differences are requisite to sharply resolve discontinuities in the traveltime derivatives, whereas centered differences improve the accuracy of the computed traveltime. A second‐order upwind essentially non‐oscillatory (ENO) scheme satisfies these requirements. It is implemented with a dynamic down ’n’ out (DNO) marching, an expanding box approach. To overcome the instability of such an expanding box scheme, the algorithm incorporates an efficient post sweeping (PS), a correction‐by‐iteration method. Near the source, an efficient and accurate mesh‐refinement initialization scheme is suggested for the DNO marching. The resulting algorithm, ENO-DNO-PS, turns out to be unconditionally stable, of second‐order accuracy, and efficient; for various synthetic and real velocity models having large contrasts, two PS iterations produce traveltimes accurate enough to complete the computation.

In this paper, we consider parallel iterative algorithms for solving the Helmholtz wave equation employing nonoverlapping domain decomposition techniques. A modified Robin interface condition incorporated with an iteration parameter is used to communicate the data near the interfaces. An automatic and non-expensive strategy for finding efficient iteration parameters is discussed in detail. Numerical results carried out on an nCUBE2 are given to demonstrate the effectiveness of the method.

This article is concerned with a level set segmentation algorithm which hybridizes gradient-based methods and the Mumford-Shah (gradient-free) method, for an efficient and reliable segmentation. We introduce a new strategy for the complementary functions uplusmn , which is computed such that the difference between their average and the given image are able to introduce a reliable driving force for the evolution of the level set function. An effective method of background subtraction is suggested in order to improve reliability of the new model. An incomplete (linearized) alternating direction implicit (ADI) method is applied for an efficient time-stepping procedure. For a fast convergence, we also suggest effective initialization strategies for the level set function. The resulting algorithm has proved to locate the desired edges satisfactorily in 2-4 ADI iterations

This article is concerned with new strategies with which explicit time-stepping procedures of PDE-based restoration models converge with a similar efficiency to implicit algorithms. Conventional explicit algorithms often require hundreds of iterations to converge. In order to overcome the difficulty and to further improve image quality, the article introduces new spatially variable constraint term and timestep size, as a method of nonflat time evolution (MONTE). It has been verified that the explicit time-stepping scheme incorporating MONTE converges in only 4-15 iterations for all restoration examples we have tested. It has proved more effective than the additive operator splitting (AOS) method in both computation time and image quality (measured in PSNR), for most cases. Since the explicit MONTE procedure is efficient in computer memory, requiring only twice the image size, it can be applied particularly for huge data sets with a great efficiency in computer memory as well.