A good result of infrared and visible image fusion should not only maintain significant contrast for distinguishing targets from the backgrounds, but also contain rich scene textures to cater human visual perception. However, previous methods usually do fully utilize information, hence their fused results sacrifice either salience thermal or sharpness textures. To address this challenge, we propose a novel Generative Adversarial Network with Full-scale skip connection dual Markovian...

Most recent video super-resolution (SR) methods either adopt an iterative manner to deal with low-resolution (LR) frames from a temporally sliding window, or leverage the previously estimated SR output help reconstruct current frame recurrently. A few studies try combine these two structures form hybrid framework but have failed give full play it. In this paper, we propose omniscient not only utilize preceding output, also outputs present and future. The is more generic because iterative,...

Pansharpening aims to fuse a multispectral (MS) image with low spatial resolution and panchromatic (PAN) high-spatial produce an both high spectral resolution. In this study, we propose variational pansharpening method by exploiting cartoon-texture similarities. After decomposition of the PAN image, cartoon component always contains global structure information, while texture includes locally patterned information. This enables that fused MS can preserve local details (e.g., high-order...

Contemporary image fusion methods face challenges in meeting the demands of dim nighttime environments, often accompanied by concealment details dark regions. In this paper, we introduce a novel approach, named LENFusion, which achieves beneficial interaction between low-light enhancement and form feedback loop. LENFusion is primarily divided into three components: Luminance Adjustment Network (LAN), Re-enhancement Fusion (RFN), Feedback (LFN). The performed two stages. initial stage, LAN...

Robust small target detection is one of the key techniques in IR search and tracking systems for self-defense or attacks. In this paper we present a robust solution single image. The ideas proposed method are to use directional support value Gaussian transform (DSVoGT) enhance targets, multiscale representation provided by DSVoGT reduce false alarm rate. original image decomposed into sub-bands different orientations convolving with filters, which deduced from weighted mapped...

As a fundamental and critical task in feature-based remote sensing image registration, feature matching refers to establishing reliable point correspondences from two images of the same scene. In this article, we propose simple yet efficient method termed linear adaptive filtering (LAF) for both rigid nonrigid apply it registration task. Our algorithm starts with putative based on local descriptors then focuses removing outliers using geometrical consistency priori together denoising theory....

In this paper, a novel self-supervised mask-optimization model, termed as SMFuse, is proposed for multi-focus image fusion. our given two source images, fully end-to-end Mask-Generator trained to directly generate the binary mask without requiring any patch operation or postprocessing through learning. On one hand, based on principle of repeated blur, we design Guided-Block with guided filter obtain an initial from narrowing solution domain and speeding up convergence generation, which...

Ultra-high performance concrete (UHPC) shows superior mechanical performance, which leads to increasing applications in infrastructure constructions that are subjected different loading (i.e., flexure, tension, compression, etc.) and environmental conditions (ambient, freeze, etc.). Among them, the flexural of UHPC under low temperatures sub-zero temperature) is still little understood especially fatigue loading. To investigate properties temperatures, eleven prisms cyclic bending at stress...

By exploiting the gradient similarity between multispectral (MS) and panchromatic (PAN) images, a variational pansharpening method based on sparse representation is proposed, observation that gradients of corresponding MS PAN images with different resolutions have similar coefficients under certain specific dictionaries. adding data fidelity term to preserve spectral information, an optimization model constructed as minimization problem energy function. The can be solved by descent...

In this study, we propose an interpretable deep network for variational pansharpening (VP), named VP-Net. Different from traditional priors using linear operators, such as the gradient, construct a prior based on similarity between panchromatic (PAN) and high-resolution multispectral (HRMS) images by nonlinear operator that can be learned through network. Considering spectral difference of various satellite (MS) imaging platforms, specifically seek aforementioned PAN image intensity HRMS to...

The existing face recognition datasets usually lack occlusion samples, which hinders the development of recognition. Especially during COVID-19 coronavirus epidemic, wearing a mask has become an effective means preventing virus spread. Traditional CNN-based models trained on are almost ineffective for heavy occlusion. To this end, we pioneer simulated dataset. In particular, first collect variety glasses and masks as occlusion, randomly combine attributes (occlusion objects, textures,and...

The existing occlusion face recognition algorithms almost tend to pay more attention the visible facial components. However, these models are limited because they heavily rely on segmentation approaches locate occlusions, which is extremely sensitive performance of mask learning. To tackle this issue, we propose a joint and identification feature learning framework for end-to-end recognition. More particularly, unlike employing an external model occlusion, design prediction module supervised...

This brief paper proposes an efficient multi-input/multi-output VLSI architecture (MIMOA) for two-dimensional lifting-based discrete wavelet transform (DWT). The novelty is the simplicity and generality to construct MIMOA, which a high-speed with computing time as low N <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> /M × image controlled increase of hardware cost. M throughput rate.

In this paper, we propose a new deep network architecture named boosting denoising net (DBDnet) for image denoising. It is residual learning that can generate noise map from noisy observation. detail, it first generates coarse via simple structure, and then updates the gradually function. The motivation of our DBDnet stems observation recovered by any algorithm cannot ideally equal ground-truth map, which typically contains noise. We call NoN, <italic...

This paper introduces a robust and scalable Gaussian process regression (GPR) model via variational learning. enables the application of processes to wide range real data, which are often large-scale contaminated by outliers. Towards this end, we employ mixture likelihood where outliers assumed be sampled from uniform distribution. We next derive formulation that jointly infers mode i.e., inlier or outlier, as well hyperparameters maximizing lower bound true log marginal likelihood. Compared...

Simultaneously fusing hyperspectral (HS), multispectral (MS), and panchromatic (PAN) images brings a new paradigm to generate high-resolution HS (HRHS) image. In this study, we propose an interpretable model-driven deep network for HS, MS, PAN image fusion, called HMPNet. We first fusion model that utilizes before describing the complicated relationship between HRHS owing their large resolution difference. Consequently, difficulty of traditional model-based approaches in designing suitable...