Journal of Visual Communication and Image Representation
Adaptive shock filter for image super-resolution and enhancement
Introduction
The goal of image super-resolution is to construct a high resolution (HR) image from a single frame or multiple low resolution (LR) images. Recently, it has become the focus of the study in many applications, such as remote sensing, military, medical imaging, public security and 3D image [1]. After a brief review on image super-resolution (SR), the SR methods are mainly divided into four categories [2]. The interpolation methods produce the HR images by using the basis function of smooth curve or curved surface. The reconstruction-based methods construct the HR image by minimizing the regularized cost function with image priors or constraints, which are usually involved with the motion estimation and registration of multi-frame images [3]. The example-based methods exploit the relationship between the HR and LR image patches for super-resolution reconstruction, or magnify the LR images by the interpolation method and then enhance the interpolated results for image super-resolution. This paper focuses on the study of single frame super-resolution based on image interpolation and enhancement.
The interpolation methods can be further divided into two categories: linear methods and nonlinear methods. The typical linear methods include nearest neighbor interpolation, the bilinear interpolation, and double three interpolation methods. These interpolation methods have high efficiency, but they tend to cause jaggy effect or loss of texture details and results in blurred images. Nonlinear interpolation methods can overcome the above shortcomings to a certain extent, e.g., the edge directing interpolation [4], the statistical interpolation [5], and the outline template interpolation [6]. However, these interpolation methods based on the local information of LR images cause the diffusion of edges and lead to blurred image and the interpolated noise.
In recent years, the partial differential equations(PDE) have been widely used in image restoration and image enhancement [7], [8], [9], [10]. The initialization of the PDE-based image super-resolution is obtained from the LR image by the interpolation methods. After solving the PDE equation, the HR image is obtained by image denoising and enhancement. Osher and Rudin [11] proposed a shock filtering model for image sharpening and enhancement. But this method also amplifies the noise. The module has good results for clean images. But it cannot remove some kind of noise, e.g., “salt and pepper” noise. Then if we have a noisy signal, it produces a lot of spurious shocks due to the influence of noise. Alvarez and Mazorra [12] introduced the Gaussian kernel function and the forward diffusion for noise reduction of image enhancement. However, it also produces fake edges, jaggy artifacts and blocking effects. The enhanced images have discontinuity edges and some isolated blocking effects. Kornprobst and Deriche [13] tuned the thresholding value to adjust the weight of shock filtering and forward diffusion. But it also causes the blurring and jaggy artifacts due to the discontinuities of the weight. Gilboa et al. [14] proposed a complex-domain shock filtering and diffusion process for image enhancement and denoising, where the imaginary part was served as an edge detector. This method can avoid the blurring artifacts, but it also causes loss of abundant details. Fu et al. [15], [16] proposed a region-based shock filtering for adaptive image enhancement. The artificial effects are decreased by use of continuous weight. But the coefficient of weight is determined by the second order directional derivative of images. The jaggy artifacts appear at certain image edges where the gradient changes fast. Subsequently, Bettahar and Stambouli [17], [18] proposed an image denoising and sharpening scheme based on curvature diffusion and shock filter. But its drawback is too complex with heavy computation.
It is found that the super-resolution methods based on interpolation have artificial defects, such as diffusion of edges, blocking effect, sawtooth effect and the interpolated noise. To solve these problems, a novel image enhancement algorithm based on adaptive shock filtering (ASF) model is proposed for image super-resolution. The weight of each shock filtering is determined by image gradients. By use of hyperbolic tangent function, the proposed algorithm can reduce the diffusion of image edges and the noise effectively. Moreover, our proposed algorithm has less sawtooth effects and blocking artifacts and can recover more fine structures than the traditional shock filter.
Section snippets
Related work
The image super-resolution methods based on interpolation usually lead to image edge diffusion, image blur and definition reduction. Since the appropriate restoration and enhancement methods can improve the resolution of images, we propose a united framework of image interpolation and adaptive shock filter for super-resolution reconstruction (See Fig. 1). The LR image was first magnified to be the initial value of the target HR image by the interpolation method. Then the adaptive shock filter
Proposed method
To solve the problem of the models described above, this paper proposes an optimized solution for adaptive shock filtering model on the basis of above analysis. The weight of shock filter is self-adaption control according to image gradients and it is continuous. It can prevent the same process for different image edges, which effectively reduces image artifacts. The final image is more consistent with the characteristics of human visual perception.
Experimental results
The proposed algorithm was compared with the popular methods published recently [12], [13], [14], [15] to verify its performance. The interpolation algorithm of contour template [6] is used to generate the initial high resolution image. For a fair comparison, these methods are tested with Matlab 2012a on the platform: 3.3 GHz dual-core CPU and 4 Gbytes RAM. The test images are selected from the standard images [19], [20]. These images are down-sampled to 1/16 of the original size. The
Conclusions
In this paper, we addressed the problem of image super-resolution and enhancement based on adaptive shock filtering model. First, the interpolation method is used to obtain the initial HR image, then the image gradient is used to adjust adaptively the weight of shock filter term, which reduces the diffusion of image edges. The forward diffusion is employed to remove noise and sharpen the image edges. Finally, the HR image is reconstructed by combining the shock filter with forward diffusion.
Acknowledgements
This work was supported by National Natural Science Foundation of China (Grant Nos. 61471272, 61201442), State Scholarship Fund of China, Natural Science Basis Research Plan in Shaanxi Province of China (Grant No. 2016JQ6068), and China Postdoctoral Science Foundation (Grant No. 2013M530481).
References (28)
Image super-resolution survey
Image Vis. Comput.
(2006)- et al.
Shock filter coupled to curvature diffusion for image denoising and sharpening
Image Vis. Comput.
(2008) - et al.
Multi-focus image fusion based on depth extraction with inhomogeneous diffusion equation
Signal Process.
(2016) - et al.
Motion analysis for image enhancement: Resolution, occlusion, and transparency
J. Vis. Commun. Image Represent.
(1993) - et al.
Joint super resolution and denoising from a single depth image
IEEE Trans. Multimedia
(2015) - et al.
Image super-resolution based on structure-modulated sparse representation
IEEE Trans. Image Process.
(2015) - et al.
New edge-directed interpolation
IEEE Trans. Image Process.
(2001) Image upsampling via imposed edge statistics
ACM Trans. Graph.
(2007)Contour stencils: total variation along curves for adaptive image interpolation
SIAM J. Imag. Sci.
(2011)- et al.
Smart interpolation by anisotropic diffusion
Edge-preserving single image superresolution
Fast image enhancement based on color space fusion
Color Res. Appl.
Adaptative shock filter for image characters enhancement and denoising
Feature oriented image enhancement using shock filters
SIAM J. Numer. Anal.
Cited by (26)
A fractional coupled system for simultaneous image denoising and deblurring
2022, Computers and Mathematics with ApplicationsSuper-resolution reconstruction of knee magnetic resonance imaging based on deep learning
2020, Computer Methods and Programs in BiomedicineCitation Excerpt :Therefore, using super resolution to improve the image quality is of great significance for the analysis and treatment of disease, especially in the early detection of knee osteoarthritis [3]. Super-resolution reconstruction technology [4,5] is a technique for reconstructing high-resolution images using single or multiple (sequence) low-resolution images. It is divided into four main approaches: interpolation [6], re-construction [7], enhancement [8], and learning [9].
Adaptive shock-diffusion model for restoration of degraded document images
2020, Applied Mathematical ModellingCitation Excerpt :This is a stable PDE model that preserves L∞ norm of the image, and so it can be associated with efficient numerical schemes. In the literature (e.g., [19,20]), some researchers have developed some PDE models by combining shock filters with nonlinear diffusion for image denoising, restoration, enhancement and super-resolution. In this section, we present an adaptive enhancement model with selective smoothing for restoration of degraded document images, involving an adaptive shock filter and a nonlinear isotropic diffusion process.
A Fuzzy Sharpness Metric for Magnetic Resonance Images
2018, Journal of Computational ScienceCitation Excerpt :In computerized analysis of Magnetic Resonance (MR) images, automated localization of structures is possible only if they have well-defined and sharp boundaries [1–3]. State of art techniques for image sharpening includes Unsharp Masking (UM) [4], shock filters [5] etc. Development of customized sharpening techniques suitable for MR images [6–8] is one of the hot areas of research in medical image computing.
A compendious study of super-resolution techniques by single image
2018, OptikCitation Excerpt :This enhances the picture quality of the output HR image. Methods based on filters [42] are also tried in interpolation to increase its performances. To overcome the shortcomings of interpolated image and its artifacts researchers applied different regularization (also called as reconstruction) based techniques on the LR image for HR image reconstruction.
This paper has been recommended for acceptance by M.T. Sun.