Super-resolution enhancement algorithms are used to estimate a high-resolution video still (HRVS) from several low-resolution frames, provided that objects within the digital image sequence move with subpixel increments. A Bayesian multi_frame algorithm is presented compute an HRVS using spatial information present each frame as well temporal due object motion between frames. However, required subpixel-resolution vectors must be estimated and noisy resulting in inaccurate field which can...

Self-learning super-resolution (SLSR) algorithms have the advantage of being independent an external training database. This paper proposes SLSR algorithm that uses convolutional principal component analysis (CPCA) and random matching. The technologies CPCA matching greatly improve efficiency self-learning. There are two main steps in this algorithm: forming testing data sets patch In set step, we propose to extract low-dimensional features set. a method quickly (PCA) each image every image....

While the performance of deep convolutional neural networks for image super-resolution (SR) has improved significantly, rapid increase memory and computation requirements hinders their deployment on resource-constrained devices. Quantized networks, especially binary (BNN) SR have been proposed to significantly improve model inference efficiency but suffer from large degradation. We observe activation distribution demonstrates very pixel-to-pixel, channel-to-channel, image-to-image variation,...

We present a nonlocal variational model for saliency detection from still images, which various features visual attention can be detected by minimizing the energy functional. The associated Euler-Lagrange equation is nonlocal<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1"><mml:mrow><mml:mi>p</mml:mi></mml:mrow></mml:math>-Laplacian type diffusion with two reaction terms, and it nonlinear diffusion. main advantage of our method that provides flexible intuitive control over...

Multiframe resolution enhancement algorithms are used to estimate a high-resolution video still (HRVS) from several low-resolution frames, provided that objects within the image sequence move with subpixel increments. A Bayesian multiframe algorithm is presented compute an HRVS using spatial information present each frame as well temporal due object motion between frames. However, required subpixel- vectors must be estimated low- and noisy resulting in inaccurate field which can adversely...

Abstract A feature-dependent variational level set formulation is proposed for image segmentation. Two second order directional derivatives act as the external constraint in evolution, with derivative across features direction playing a key role contour extraction and another only slightly contributes. To overcome local gradient limit, we integrate information from maximal (in magnitude) second-order into common framework. It naturally encourages function to deform (up or down) opposite...

From the point of view product quality, considering performance characteristics laminated elastomers, this paper proposes a classification method for elastomers based on K-medoids algorithm. In addition, author also analyzed and verified feasibility beneficial effects method. The analysis results show that mass proposed in is feasible. Moreover, according to method, assembling elastic element group can further improve its group. When completes same action, required force decreases...

This paper presents an image interpolation model with local and nonlocal regularization. A bounded variation (BV) regularizer is formulated by exponential function including gradient. It acts as the Perona-Malik equation. Thus our BV possesses properties of anisotropic diffusion equation functional. The total (TV) dissipates energy along orthogonal direction to gradient avoid blurring edges. derived efficiently reconstructs real image, leading a natural which reduces staircase artifacts. We...

This paper presents an isophote-oriented image interpolation method that attempts to produce smooth reconstructions of the image¿s level curves while still maintaining fidelity. The proposed employs kernel is obtained by looking for explicit solution a nonlinear PDE (partial differential equation). preserve local continuation isophotes (curves constant intensity). By doing so, curvature interpolated reduced, and, thus, zigzagging artifacts are largely suppressed. Moreover, techniques can...

An energy functional with bidirectional flow is presented to sharpen image by reducing its edge width, which performs a forward diffusion in brighter lateral on ramp and backward that proceeds darker lateral. We first consider the equations as<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1"><mml:mrow><mml:msub><mml:mrow><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mi>2</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:math>gradient flows integral functionals then modify inner...

Image upscaling is needed in many areas. There are two types of methods: methods based on a simple hypothesis and machine learning. Most the learning‐based have disadvantages: no support provided for variety factors, training process with high time cost required, large amount storage space high‐end equipment required. To avoid disadvantages learning, images promising strategy but always produces jaggy artifacts. The authors propose new method to remove these jagged They consider an edge...

Aiming at the problem that blurred digital video is easy to lose inter-frame information and ignore spatiotemporal during restoration, a image deblurring algorithm based on denoising engine proposed. We extend adaptive Laplacian regularization term constructed by field of restoration. Firstly, self-similarity redundant can be gained through nonlocal means (NLM) regularization, then we present new restoration model mixing different regularizers, especially combining NLM regularizer with...

Image super-resolution reconstruction is an ill-posed problem in nature because there are infinitely many high-resolution images that can be reconstructed from a low-resolution image. To limit the solution space and make good use of widely existing cross-scale feature similarities natural images, we propose dual regression network based on non-local attention mechanism, which not only restrict image but also better embrace abundant external information. The method has been tested five...

In order to improve the performance of image denoising, in case can not only reduce computational cost at same time ensure superiority we made a change on basis original network, through way increasing network breadth rather than depth for more features, and running speed, by means expansion convolution extract information used denoising task. A large number experimental results show that this kind gradient explosion, but also effectively noise intensity image.