Adaptive Regularization Term Research Articles

Image Super-Resolution via Adaptive Regularization Term of Compressed Sensing

Image Super-Resolution via Adaptive Regularization Term of Compressed Sensing

Weighted side-window based gradient guided image filtering

Image filtering under guidance image, known as guided filtering (GF), has been successfully applied to a variety of applications. Existing GF methods utilize either conventional full window-based framework (FWF) or simple uniformly weighted aggregation strategy (UWA); thereby they suffer from edge-blurring. In this paper, based upon gradient guided filtering (GGF), a weighted side-window based gradient guided filtering (WSGGF) is proposed to address the aforementioned problem. First, both regression and adaptive regularization terms in GGF are improved upon eight side windows by introducing side window-based framework (SWF). L1 norm is adopted to choose the results calculated in side windows. Second, UWA strategy in GGF is replaced by a refined variance-based weighted average (VWA) aggregation. In VWA, the value of each weight is chosen inversely proportional to the corresponding estimator. We show that with these improvements our method can well retain the edge sharpness and is robust to visual artifacts. To cut down the time consumption, a fast version of WSGGF (FWSGGF) is further proposed by incorporating a simple but effective down-sampling strategy, which is about four times faster while maintaining the superior performance. By comparing with the state-of-the-art (SOTA) methods on edge-aware smoothing, detail enhancement, high dynamic range image (HDR) compression, image luminance adjustment, depth map upsampling and single image haze removal, the effectiveness and flexibility of our proposed methods are verified. The source code is available at: https://github.com/weimin581/WSGGF

Adaptive anisotropic pixel-by-pixel correction method for a space-variant degraded image.

Large field-of-view optical imaging systems often face challenges in the presence of space-variant degradation. The existence of degradation leads to target detection and recognition being difficult or even unsuccessful. To address this issue, this paper proposes an adaptive anisotropic pixel-by-pixel space-variant correction method. First, we estimated region acquisition of local space-variant point spread functions (PSFs) based on Haar wavelet degradation degree distribution, and obtained initial PSF matrix estimation with inverse distance weighted spatial interpolation. Then, we established a pixel-by-pixel space-variant correction model based on the PSF matrix. Third, we imposed adaptive sparse regularization terms of the Haar wavelet based on the adaptive anisotropic iterative reweight strategy and non-negative regularization terms as the constraint in the pixel-by-pixel space-variant correction model. Finally, as the correction process is refined to each pixel, the split-Bregman multivariate separation solution algorithm was employed for the pixel-by-pixel spare-variant correction model to estimate the final PSF matrix and the gray value of each pixel. Through this algorithm, the "whole image correction" and "block correction" is avoided, the "pixel-by-pixel correction" is realized, and the final corrected images are obtained. Experimental results show that compared with the current advanced correction methods, the proposed approach in the space-variant wide field correction of a degraded image shows better performance in preserving the image details and texture information.

Boundary-aware registration network for 4D-CT lung image with sliding motion

Background and ObjectiveFast and accurate registration of 4D-CT lung images is significant for respiratory motion modeling and radiotherapy planning. However, since the complex respiratory motion involves sliding motion at lung boundary, the traditional registration methods and regularization terms perform poorly in recovering both sliding and smooth deformation in 4D-CT lung image registration. MethodsIn order to overcome these limitations of the traditional registration methods and regularization terms, we propose a boundary-aware registration model with a spatially adaptive regularization term. We incorporate a lung segmentation network into the registration model. With the lung boundary-aware information from the segmentation network, we construct a spatially adaptive regularization term, which integrates the smooth regularization and the non-smooth regularization, to accommodate both smooth and sliding motion. ResultsWe evaluate the proposed registration model on clinical 4D-CT data and the public DIR-Lab dataset. Our model provides a minimum Target Registration Error (1.59 ± 0.57 mm) of landmarks compared with the other lung registration methods. The ablation studies show that the proposed spatially adaptive regularization term provides superior performance in HD (13.75 ± 3.36 mm) and MHD (1.63 ± 0.32 mm) to the smooth regularization term and non-smooth regularization term. ConclusionsThe proposed boundary-aware registration model enables adaptive regularization term, which can flexibly regulate both the sliding motion at the lung boundary and the smooth motion inside the lung simultaneously. Therefore, our model can perform fast and accurate registration for 4D-CT lung images with sliding motion, which is beneficial to respiratory motion modeling and lung cancer radiotherapy.

Adaptive graph regularization and self-expression for noise-aware feature selection

Many traditional unsupervised feature selection algorithms utilize manifold information to mine the local structure of the data. However, the noise existing in the raw data reduces the accuracy of the manifold information of data, which affects the learning effect of the entire algorithm. In order to solve the above problems and more fully find the internal structure inside the data, this paper proposes an adaptive graph regularization and self-expression for noise-aware feature selection (ASNFS). Firstly, the algorithm adopts non-negative matrix factorization to decompose the raw data matrix, and utilizes the low-dimensional matrix generated after decomposition to replace the raw high-dimensional data matrix. This allows the algorithm to reveal some internal structural information of the raw data while reducing the dimensionality of the data. ASNFS also introduces the orthogonal basis clustering with excellent clustering effect, and the interpretability of the algorithm is enhanced. Secondly, in addition to preserving the manifold information in the low-dimensional projection subspace, the algorithm also preserves the manifold information in the non-negative matrix factorization subspace. Meanwhile, the adaptive graph regularization term added to the objective function enables the algorithm to continuously update the similarity matrix. It can effectively remove the noise inside the raw data, and prevent the occurrence of over-fitting phenomenon of the experimental results caused by the fixed similarity matrix. Finally, the similarity matrix retains the local structure information of the data with each iteration, and the results of the feature selection are reused for the construction of the similarity matrix. ASNFS adopts an alternate iterative method to optimize the objective function, which is simple and effective. Then, the algorithm complexity and convergence are analyzed. ASNFS is compared with seven feature selection algorithms on nine datasets, and the experimental results reflect the effectiveness of ASNFS in feature selection.

Adaptive Spatial Regularization Target Tracking Algorithm Based on Multifeature Fusion

The accuracy and robustness of object-tracking algorithms are challenging tasks in the field of artificial intelligence. The discriminative correlation filter has fast tracking speed and target discrimination ability in visual target tracking, but it will be affected by unnecessary boundary effects. Although spatially regularized discriminative correlation filters (SRDCF) effectively solve this problem, it faces a slow tracking speed and cannot meet the real-time requirement. Moreover, the constraints added by this algorithm are fixed during the tracking process, which needs to reflect better the characteristics and appearance changes of a particular object. Therefore, we propose an adaptive spatial regularization target tracking model based on multifeature fusion and learn the spatial regularization weights by adaptive spatial regularization terms, so that the filter can adapt to the changes of the target, which can highlight the target area and suppress the background area more accurately. At the same time, in order to cope with the problem of target loss when the target is obscured by a large area or rotated to a large extent, texture features are introduced to improve the deficiencies of HOG features and CN features, so as to achieve the complementary advantages among different features. In addition, the benchmark algorithm SRDCF uses Gaussian–Seidel iterative method to solve the filter, which makes the tracking speed very slow and cannot be tracked in real time. Therefore, this paper proposes an ADMM (alternating direction method of multiplier) algorithm to optimize the solution of the filter and achieve the tracking real-time requirement. The experimental results on the OTB-2015 dataset show that the accuracy and success rate of the improved algorithm reach 86% and 81.9%, respectively. At the same time, in the OTB-2013 dataset, it has increased by 3.4% and 6.3%, respectively, on the basis of the benchmark algorithm. It is verified that the improved model not only improves the robustness of the algorithm but also can better adapt to the changes of the target, while effectively reducing the computational overhead and meeting the tracking real-time requirements.

An iterative data-driven turbulence modeling framework based on Reynolds stress representation

Data-driven turbulence modeling studies have reached such a stage that the basic framework is settled, but several essential issues remain that strongly affect the performance. Two problems are studied in the current research: (1) the processing of the Reynolds stress tensor and (2) the coupling method between the machine learning model and flow solver. For the Reynolds stress processing issue, we perform the theoretical derivation to extend the relevant tensor arguments of Reynolds stress. Then, the tensor representation theorem is employed to give the complete irreducible invariants and integrity basis. An adaptive regularization term is employed to enhance the representation performance. For the coupling issue, an iterative coupling framework with consistent convergence is proposed and then applied to a canonical separated flow. The results have high consistency with the direct numerical simulation true values, which proves the validity of the current approach.

Robust dual-graph regularized and minimum redundancy based on self-representation for semi-supervised feature selection

Partial labeled data is ubiquitous in the big data era. Selecting informative features, and avoiding redundant and noise features is an important task for constructing robust learning models. The inherent characteristics of samples and features need to be surveyed simultaneously. In this study, a feature selection method for partial labeled data based on dual-graph regularized is proposed. Self-representation can well mine the relation between features. Then, the self-representation is combined with the framework of sparse learning to well embody the character of data and an adaptive redundancy regularization term based on self-representation is designed to minimize the redundancy between features. In the self-representation based regularization term, the relation between features is updated dynamically during the iteration so as to capture the inherent relation between features more accurately. The manifold structure in both feature space and data space are considered jointly. Moreover, the L2,p-norm is imposed on the feature self-representation regularization term, self-representation coefficient matrix, and feature selection matrix, respectively. These constrains aim to enhance its robustness to outliers, and select the representative features with discriminative and low redundancy. The convex (p=1) and non-convex (0<p<1) are involved in the proposed method. Then, a unified solution for the proposed method is investigated and its convergence is proved. Experimental results on public data sets show that the proposed method is effective for classification tasks when compared with some feature selection methods.

Adaptive Sparse Domain Selection for Weather Radar Superresolution

Accurate and high-resolution weather radar data reflecting detailed structure information of radar echo plays an important role in analysis and forecast of extreme weather. Typically, this is done using interpolation schemes, which only use several neighboring data values for computational approximation to get the estimated, resulting the loss of intense echo information. Focus on this limitation, a superresolution reconstruction algorithm of weather radar data based on adaptive sparse domain selection (ASDS) is proposed in this article. First, the ASDS algorithm gets a compact dictionary by learning the precollected data of model weather radar echo patches. Second, the most relevant subdictionaries are adaptively select for each low-resolution echo patches during the spare coding. Third, two adaptive regularization terms are introduced to further improve the reconstruction effect of the edge and intense echo information of the radar echo. Experimental results show that the ASDS algorithm substantially outperforms interpolation methods for ×2 and ×4 reconstruction in terms of both visual quality and quantitative evaluation metrics.

Nonuniform B-Spline Lung Image Registration Based on Adaptive Regularization Term

Nonuniform B-Spline Lung Image Registration Based on Adaptive Regularization Term

An intelligent fault diagnosis method for rolling bearings based on feature transfer with improved DenseNet and joint distribution adaptation

Mechanical intelligent fault diagnosis is an important method to accurately identify the health status of mechanical equipment. Traditional fault diagnosis methods perform poorly in the diagnosis of rolling bearings under complex conditions. In this paper, a feature transfer learning model based on improved DenseNet and joint distribution adaptation (FT-IDJ) is proposed. With this model, we apply it to implement rolling bearing fault diagnosis. A lightweight DenseNet model is firstly proposed to extract the transferable features of the raw vibration signal. Furthermore, the parameters in the DenseNet are constrained by the domain adaptive regularization term and pseudo label learning. The marginal distribution discrepancy and the conditional distribution discrepancy of the learned transferable features are reduced by this way. The proposed method is validated by the diagnosis experiments with CWRU and Jiangnan University rolling bearing datasets. The experimental results showed that the proposed FT-IDJ can classify common rolling bearing faults such as inner ring fault, outer ring fault and rolling element fault. Also, it has higher classification accuracy than DAN and other eight methods, which demonstrates its effectively learning transferable features from auxiliary data.

Learning adaptive spatial regularization and aberrance repression correlation filters for visual tracking

This paper proposes a correlation filter tracking algorithm based on adaptive spatial regularization and aberrance repression aiming at the problem that the spatial regularization weight of the background-aware correlation filter is fixed and does not adapt to the change of the target, and the problem that enlarging search area may introduce background noise, decreasing the discrimination ability of filters. First, FHOG features, CN features, and gray features are extracted to enhance the algorithm's ability to express the target. Second, aberrance repression terms are added to the target function to constrain the response map of the current frame, and to enhance the filter's discrimination ability to alleviate the filter model degradation. Finally, adaptive spatial regularization terms are added to the objective function to make the spatial regularization weights being updated as the objective changes, so that the filter can make full use of the target's diversity information. This paper involves experiments on the public data sets OTB-2013, OTB-2015 and VOT2016 to evaluate the proposed algorithm. The experimental results show that the speed of the algorithm used in this paper is 20 frames/s, evaluation indicators such as distance accuracy and success rate are superior to comparison algorithms, and it has good robustness in a variety of complex scenarios such as occlusion, background interference, and rotation changes.

Adaptive active contour model driven by image data field for image segmentation with flexible initialization

In this paper, a novel adaptive active contour model based on image data field for image segmentation with robust and flexible initializations is proposed. We firstly construct a new external energy term deduced from the image data field that drives the level set function to move in the opposite direction along the boundaries of object and an adaptive length regularization term based on the image local entropy. The designed external energy and length regularization term are then incorporated into a variationlevel set framework with an additional penalizing energy term. Due to the adaptive sign–changing property of the external energy and the adaptive length regularization term, the proposed model can tackle images with clutter background and noise, the level set function can be initialized as any bounded functions (e.g., constant function), which implies the proposed model is robust to initialization of contours. Experimental results on both synthetic and real images from different modalities confirm the effectiveness and competivive performance of the proposed method compared with other representative models.

A new denoising model for multi-frame super-resolution image reconstruction

Multi-frame image super-resolution (SR) aims to combine the sub-pixel information from a sequence of low-resolution (LR) images to build a high-resolution (HR) one. SR techniques usually suffers from annoying restoration artifacts such as noise, jagged edges, and staircasing effect. In this paper, we aim to increase the performance of SR reconstitution under a variational framework using adaptive diffusion-based regularization term. We propose a new tensor based diffusion regularization that takes the benefit from the diffusion model of Perona–Malik in the flat regions and use a nonlinear tensor derived from the diffusion process of Weickert filter near boundaries. Thus, the proposed SR approach can preserve important image features (sharp edges and corners) much better while avoiding artifacts. The synthetic and real experimental results show the effectiveness of the proposed regularisation compared to other methods in both quantitatively and visually.

Hyperspectral Image Denoising with a Combined Spatial and Spectral Weighted Hyperspectral Total Variation Model

. Hyperspectral image (HSI) denoising is a prerequisite for many subsequent applications. For an HSI, the level and type of noise often vary with different bands and spatial positions, which make it difficult to effectively remove noise while preserving textures and edges. To alleviate this problem, we propose a new total-variation model. The main contribution of the proposed approach lies in that the adaptive regularization terms in both the spatial and the spectral dimensions are designed separately and then combined into a unified framework. The 2 separate regularization terms allow a better description of the intrinsic nature of the original HSI data and can simultaneously penalize the noise from both the spatial and spectral perspectives. The designed weights for the regularization terms are positively correlated with the magnitude of the noise intensity and negatively correlated with the signal variation; thus, the original signal can be accurately retained and the noise can be effectively suppressed. To efficiently process the HSI, which appears as a huge data cube, a new optimization algorithm based on the alternating direction method of multipliers (ADMM) procedure is proposed to solve the new model. Experiments using HYDICE and AVIRIS images were conducted to validate the effectiveness of the proposed method.Résumé. Hyperspectrale l’image (HSI) débruitage est une condition préalable pour de nombreuses applications ultérieures. Pour un HSI, le niveau et le type de bruit varie souvent avec différents groupes et positions spatiales, ce qui rend difficile d’éliminer efficacement le bruit tout en préservant les textures et les bords. Pour pallier ce problème, nous proposons un nouveau modèle de variation totale. Les principales contributions de l’approche proposée mensonge dans la conception des termes de régularisation adaptative dans les deux dimensions spatiales et spectrales, et en les combinant dans un cadre unifié. Les deux termes de régularisation séparés permettent une meilleure description de la nature intrinsèque des données HSI original et peuvent pénaliser simultanément le bruit à la fois des perspectives spatiales et spectrales. Les poids conçus pour les termes de régularisation sont en corrélation positive avec la grandeur de l’intensité du bruit et corrélation négative avec la variation de signal; ainsi, le signal d’origine peut être retenu avec précision et le bruit peut être efficacement supprimée. Pour traiter efficacement le HSI, qui apparaît comme un énorme cube de données, un nouvel algorithme d’optimisation basé sur la méthode de direction alternée de multiplicateurs «alternating direction method of multipliers» (ADMM) procédure est proposée pour résoudre le nouveau modèle. Des expériences utilisant des images AVIRIS et HYDICE et ont été menées afin de valider l’efficacité de la méthode proposée.

MO‐G‐17A‐05: PET Image Deblurring Using Adaptive Dictionary Learning

Purpose:The aim of this work is to deblur PET images while suppressing Poisson noise effects using adaptive dictionary learning (DL) techniques.Methods:The model that relates a blurred and noisy PET image to the desired image is described as a linear transform y=Hm+n where m is the desired image, H is a blur kernel, n is Poisson noise and y is the blurred image. The approach we follow to recover m involves the sparse representation of y over a learned dictionary, since the image has lots of repeated patterns, edges, textures and smooth regions. The recovery is based on an optimization of a cost function having four major terms: adaptive dictionary learning term, sparsity term, regularization term, and MLEM Poisson noise estimation term. The optimization is solved by a variable splitting method that introduces additional variables. We simulated a 128×128 Hoffman brain PET image (baseline) with varying kernel types and sizes (Gaussian 9×9, σ=5.4mm; Uniform 5×5, σ=2.9mm) with additive Poisson noise (Blurred). Image recovery was performed once when the kernel type was included in the model optimization and once with the model blinded to kernel type. The recovered image was compared to the baseline as well as another recovery algorithm PIDSPLIT+ (Setzer et. al.) by calculating PSNR (Peak SNR) and normalized average differences in pixel intensities (NADPI) of line profiles across the images.Results:For known kernel types, the PSNR of the Gaussian (Uniform) was 28.73 (25.1) and 25.18 (23.4) for DL and PIDSPLIT+ respectively. For blinded deblurring the PSNRs were 25.32 and 22.86 for DL and PIDSPLIT+ respectively. NADPI between baseline and DL, and baseline and blurred for the Gaussian kernel was 2.5 and 10.8 respectively.Conclusion:PET image deblurring using dictionary learning seems to be a good approach to restore image resolution in presence of Poisson noise. GE Health Care

Super-Resolution Reconstruction based on Tukey Norm and Adaptive Bilateral Total Variation

In Bilateral Total Variation (BTV) regularized super-resolution reconstruction (SRR), the fidelity item is only applicable to a specific noise model, and the fixed weight of BTV regularization term cannot adapt to the changes in an image. Thus, this paper proposes a SRR algorithm based on the Tukey fidelity term and adaptive BTV regularization term. The Tukey fidelity term has a more effective outliers suppression feature to deal with complex noises, and the weight of adaptive BTV regularization term can resize itself according to the changes of image textures, which can achieve the purposes of suppressing noises and preserving edges. Experimental results show that, compared with other algorithms, the proposed algorithm has better vision effects and higher Peak Signal-to-noise Ratio (PSNR) values.

Adaptive variational models for image decomposition

Image decomposition refers to the splitting of an image into two or more components. In this paper, a clean image is separated into two parts: one is the cartoon component, consisting only of geometric structure, and the other is the oscillatory component, consisting of texture. Three parts for noisy image are considered: cartoon, texture, and noise. To better decompose an image, we propose two new variational models. In our models, two adaptive regularization terms are introduced. The two regularization terms are determined by an adaptive function which can discriminate the cartoon and texture of an image automatically. Experimental results illustrate the effectiveness of the proposed models for image decomposition.

Hyperspectral imagery super-resolution by sparse representation and spectral regularization

For the instrument limitation and imperfect imaging optics, it is difficult to acquire high spatial resolution hyperspectral imagery. Low spatial resolution will result in a lot of mixed pixels and greatly degrade the detection and recognition performance, affect the related application in civil and military fields. As a powerful statistical image modeling technique, sparse representation can be utilized to analyze the hyperspectral image efficiently. Hyperspectral imagery is intrinsically sparse in spatial and spectral domains, and image super-resolution quality largely depends on whether the prior knowledge is utilized properly. In this article, we propose a novel hyperspectral imagery super-resolution method by utilizing the sparse representation and spectral mixing model. Based on the sparse representation model and hyperspectral image acquisition process model, small patches of hyperspectral observations from different wavelengths can be represented as weighted linear combinations of a small number of atoms in pre-trained dictionary. Then super-resolution is treated as a least squares problem with sparse constraints. To maintain the spectral consistency, we further introduce an adaptive regularization terms into the sparse representation framework by combining the linear spectrum mixing model. Extensive experiments validate that the proposed method achieves much better results.

Image Deblurring and Super-Resolution by Adaptive Sparse Domain Selection and Adaptive Regularization

As a powerful statistical image modeling technique, sparse representation has been successfully used in various image restoration applications. The success of sparse representation owes to the development of the l(1)-norm optimization techniques and the fact that natural images are intrinsically sparse in some domains. The image restoration quality largely depends on whether the employed sparse domain can represent well the underlying image. Considering that the contents can vary significantly across different images or different patches in a single image, we propose to learn various sets of bases from a precollected dataset of example image patches, and then, for a given patch to be processed, one set of bases are adaptively selected to characterize the local sparse domain. We further introduce two adaptive regularization terms into the sparse representation framework. First, a set of autoregressive (AR) models are learned from the dataset of example image patches. The best fitted AR models to a given patch are adaptively selected to regularize the image local structures. Second, the image nonlocal self-similarity is introduced as another regularization term. In addition, the sparsity regularization parameter is adaptively estimated for better image restoration performance. Extensive experiments on image deblurring and super-resolution validate that by using adaptive sparse domain selection and adaptive regularization, the proposed method achieves much better results than many state-of-the-art algorithms in terms of both PSNR and visual perception.