Patch-based Sparse Representation Research Articles

Joint Sparse Representation-based Single Image Super-Resolution for Remote Sensing Applications

Sparse representation-based single image super-resolution (SISR) methods use a coupled overcomplete dictionary trained from high-resolution (HR) images/ image patches. Since, remote sensing (RS) satellites capture images of large areas, these images usually have poor spatial resolution and obtaining an effective dictionary as such would be very challenging. Moreover, traditional patch-based sparse representation (PSR) models for reconstruction tend to give unstable sparse solution and produce visual artefact in the recovered images. To mitigate these problems, in this paper, we have proposed an adaptive joint sparse representation (JSR)-based SISR method that is dependent only on the input low-resolution (LR) image for dictionary training and sparse reconstruction. The new model combines patch-based local sparsity and group sparse representation (GSR)-based non-local sparsity in a single framework, which helps in stabilizing the sparse solution and improve the SISR results. Experimental results are evaluated both visually and quantitatively for several RGB and multispectral RS datasets, where the proposed method shows improvements in PSNR by 1–4 dB and 2–3 dB over the state-of-the-art sparse representation- and deep learning-based SR methods, respectively. Land cover classification applied on the super-resolved images further validate the advantages of the proposed method. Finally, for practical RS applications, we have performed parallel implementation in general purpose graphics processing units (GPGPU) and achieved significant speed-ups (30-40×) in the execution time.

A Hybrid Sparse Representation Model for Image Restoration.

Group-based sparse representation (GSR) uses image nonlocal self-similarity (NSS) prior to grouping similar image patches, and then performs sparse representation. However, the traditional GSR model restores the image by training degraded images, which leads to the inevitable over-fitting of the data in the training model, resulting in poor image restoration results. In this paper, we propose a new hybrid sparse representation model (HSR) for image restoration. The proposed HSR model is improved in two aspects. On the one hand, the proposed HSR model exploits the NSS priors of both degraded images and external image datasets, making the model complementary in feature space and the plane. On the other hand, we introduce a joint sparse representation model to make better use of local sparsity and NSS characteristics of the images. This joint model integrates the patch-based sparse representation (PSR) model and GSR model, while retaining the advantages of the GSR model and the PSR model, so that the sparse representation model is unified. Extensive experimental results show that the proposed hybrid model outperforms several existing image recovery algorithms in both objective and subjective evaluations.

Intracardiac Mass Detection and Classification Using Double Convolutional Neural Network Classifier

Intracardiac masses identification in the images of echocardiogram images in one of the most essential tasks in making the diagnosis of cardiac disease. For making the improvement in accuracy over the diagnosis as a new complete method of classifying the echocardiogram images automatically which is based on robust back propagation neural network algorithm in being proposed for distinguishing intracardiac thrombi and tumor. Initially, the cropping over the specific region is done in order to make the definition of the mass area. Later on, as the second step the processing of globally unique denoising technique is being implied for the removal of speckle and in order to make the preservation of anatomical structured component in the image. This is defined in terms of preprocessing and it is carried out by Patch-based sparse representation. Subsequently the description of the mass contour and its interconnected wall of the artery are being done by the segmentation mechanism denoted as Linear Iterative Vessel Segmentation model. As the prefinal stage, the processing of boundary, texture and the motion features are being carried out through the processing by double convolutional neural network (DCNN) classifier in order to determine the classification of two different masses. Totally 108 cardiac masses images are being collected for accessing the effectiveness of the classifier. It is also realized with the various state of the art classifiers as projected the demonstration of the greatest performance that has been disclosed with an achievement of 98.98% of accuracy, 98.89% of sensitivity and 99.16% of specificity that has been resulted for DCNN classifier. It determines the explication that the proposed method is capable of performing the classification of intracardiac thrombi and tumors in the echocardiography and ensures for potentially assisting the medical doctors who are in the clinical practice.

Image Restoration Using Joint Patch-Group-Based Sparse Representation

Sparse representation has achieved great success in various image processing and computer vision tasks. For image processing, typical patch-based sparse representation (PSR) models usually tend to generate undesirable visual artifacts, while group-based sparse representation (GSR) models lean to produce over-smooth effects. In this paper, we propose a new sparse representation model, termed joint patch-group based sparse representation (JPG-SR). Compared with existing sparse representation models, the proposed JPG-SR provides an effective mechanism to integrate the local sparsity and nonlocal self-similarity of images. We then apply the proposed JPG-SR to image restoration tasks, including image inpainting and image deblocking. An iterative algorithm based on the alternating direction method of multipliers (ADMM) framework is developed to solve the proposed JPG-SR based image restoration problems. Experimental results demonstrate that the proposed JPG-SR is effective and outperforms many state-of-the-art methods in both objective and perceptual quality.

Group-Based Sparse Representation Based on lp -Norm Minimization for Image Inpainting

As a powerful statistical image modeling technique, sparse representation has been successfully applied in various image restoration applications. Most traditional methods depend on ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> -norm optimization and patch-based sparse representation models. However, these methods have two limits: high computational complexity and the lack of the relationship among patches. To solve the above problems, we choose the group-based sparse representation models to simplify the computing process and realize the nonlocal self-similarity of images by designing the adaptive dictionary. Meanwhile, we utilize Ipnorm minimization to solve nonconvex optimization problems based on the weighted Schatten p-norm minimization, which can make the optimization model more flexible. Experimental results on image inpainting show that the proposed method has a better performance than many current state-of-the-art schemes, which are based on the pixel, patch, and group respectively, in both peak signal-to-noise ratio and visual perception.

Axis-guided patch based accurate segmentation for pathological vessels using adaptive weight sparse representation

Background and objectiveOver decades of research and development in vessel segmentation, accurate and reliable methods targeting at pathological vessels are few. Addressing this challenge, we try to delineate the vessel boundary accurately with the presence of pathologies. MethodsA novel segmentation framework is presented in this work, a vessel axis tracking algorithm + a patch-based sparse representation. The patch-based algorithm is navigated by the vessel axis tracking algorithm. Within the training process, multi-scale training samples have been used, which has the potential to be both physiologically accurate and computationally effective. The redundant information embedded in the multi-scale training samples is employed to delineate the pathological vessel accurately. To further reduce the computational burden caused by the patch-based sparse representation, a multi-scale dictionary has been generated, and adaptive weights have been assigned to the scale-specific sub-dictionary atoms, for computational efficiency. ResultsOur method is evaluated by comparing with two state-of-the-art methods, on synthetic complex-structured datasets and real clinical datasets. The performance of our method is promising over the evaluation, since the overlap ratios of our method are high over all the datasets, around 91%, much better than two state-of-the-art methods.

Similarity mapping for robust face recognition via a single training sample per person

Face recognition using a single training sample per person (STSPP) is very challenging due to significant variations in the query sample. Although most of sparse representation based methods eliminate the variations to obtain discriminative power, their recognition performance usually degrades as the number of training samples per person decreases. To improve recognition performance on the STSPP problem, in this paper, we propose a novel face recognition method: generating new training samples by preserving the variations of each query sample onto training samples instead of eliminating the variations, and then identifying the query one via the most similar generated training sample. Specifically, we formulate face recognition as an patch-based sparse representation problem, and then efficiently solve it by dividing the optimization procedure into three individual sub-problems: selecting training patches, weighting selected patches and calculating residuals. Meanwhile, we interpret the proposed algorithm from a probability modeling viewpoint. Experiments on multiple benchmark databases demonstrate its superior recognition performance to state-of-the-art methods on faces under illumination changes, occlusions and facial expressions, with lower test time costs.

Construction of 4D infant cortical surface atlases with sharp folding patterns via spherical patch‐based group‐wise sparse representation

4D (spatial + temporal) infant cortical surface atlases covering dense time points are highly needed for understanding dynamic early brain development. In this article, we construct a set of 4D infant cortical surface atlases with longitudinally consistent and sharp cortical attribute patterns at 11 time points in the first six postnatal years, that is, at 1, 3, 6, 9, 12, 18, 24, 36, 48, 60, and 72 months of age, which is targeted for better normalization of the dynamic changing early brain cortical surfaces. To ensure longitudinal consistency and unbiasedness, we adopt a two-stage group-wise surface registration. To preserve sharp cortical attribute patterns on the atlas, instead of simply averaging over the coregistered cortical surfaces, we leverage a spherical patch-based sparse representation using the augmented dictionary to overcome the potential registration errors. Our atlases provide not only geometric attributes of the cortical folding, but also cortical thickness and myelin content. Therefore, to address the consistency across different cortical attributes on the atlas, instead of sparsely representing each attribute independently, we jointly represent all cortical attributes with a group-wise sparsity constraint. In addition, to further facilitate region-based analysis using our atlases, we have also provided two widely used parcellations, that is, FreeSurfer parcellation and multimodal parcellation, on our 4D infant cortical surface atlases. Compared to cortical surface atlases constructed with other methods, our cortical surface atlases preserve sharper cortical folding attribute patterns, thus leading to better accuracy in registration of individual infant cortical surfaces to the atlas.

A DICTIONARY LEARNING APPROACH FOR FRACTAL IMAGE CODING

In recent years, sparse representations of images have shown to be efficient approaches for image recovery. Following this idea, this paper investigates incorporating a dictionary learning approach into fractal image coding, which leads to a new model containing three terms: a patch-based sparse representation prior over a learned dictionary, a quadratic term measuring the closeness of the underlying image to a fractal image, and a data-fidelity term capturing the statistics of Gaussian noise. After the dictionary is learned, the resulting optimization problem with fractal coding can be solved effectively. The new method can not only efficiently recover noisy images, but also admirably achieve fractal image noiseless coding/compression. Experimental results suggest that in terms of visual quality, peak-signal-to-noise ratio, structural similarity index and mean absolute error, the proposed method significantly outperforms the state-of-the-art methods.

Group Sparse Representation Based Dictionary Learning for SAR Image Despeckling

Since the sparse representation coefficients of synthetic aperture radar (SAR) images often appear in clusters with intrinsic structure, traditional sparse representation theory cannot capture this property. In this paper, the concept of group sparse representation (GSR) is utilized to exploit the intrinsic structure of SAR images. Different from traditional patch-based sparse representation theory, GSR is able to sparsely represent images in the domain of group which contains the image patches with similar structure. Based on the multiplicative speckle noise model, a novel dictionary learning algorithm based on GSR (GSR-DL) for SAR image despeckling is proposed. The proposed algorithm mainly consists of three steps. First, in order to realize the recovery of despeckled SAR image by the GSR model, a mean filter is included in the modeling process. Second, the proposed GSR-DL algorithm is used to calculate the optimal dictionary and group sparse representation coefficients. Third, the despeckled SAR image is reconstructed by the learned dictionary and coefficients. The experimental results on SAR images manifest that the proposed GSR-DL algorithm achieves a better performance than other state-of-the-art despeckling algorithms.

Speckle Reduction on Ultrasound Liver Images Based on a Sparse Representation over a Learned Dictionary

Ultrasound images are corrupted with multiplicative noise known as speckle, which reduces the effectiveness of image processing and hampers interpretation. This paper proposes a multiplicative speckle suppression technique for ultrasound liver images, based on a new signal reconstruction model known as sparse representation (SR) over dictionary learning. In the proposed technique, the non-uniform multiplicative signal is first converted into additive noise using an enhanced homomorphic filter. This is followed by pixel-based total variation (TV) regularization and patch-based SR over a dictionary trained using K-singular value decomposition (KSVD). Finally, the split Bregman algorithm is used to solve the optimization problem and estimate the de-speckled image. The simulations performed on both synthetic and clinical ultrasound images for speckle reduction, the proposed technique achieved peak signal-to-noise ratios of 35.537 dB for the dictionary trained on noisy image patches and 35.033 dB for the dictionary trained using a set of reference ultrasound image patches. Further, the evaluation results show that the proposed method performs better than other state-of-the-art denoising algorithms in terms of both peak signal-to-noise ratio and subjective visual quality assessment.

CONSTRUCTION OF SPATIOTEMPORAL NEONATAL CORTICAL SURFACE ATLASES USING A LARGE-SCALE DATASET.

The cortical surface atlases constructed from a large representative population of neonates are highly needed in the neonatal neuroimaging studies. However, existing neonatal cortical surface atlases are typically constructed from small datasets, e.g., tens of subjects, which are inherently biased and thus are not representative to the neonatal population. In this paper, we construct neonatal cortical surface atlases based on a large-scale dataset with 764 subjects. To better characterize the dynamic cortical development during the first postnatal weeks, instead of constructing just a single atlas, we construct a set of spatiotemporal atlases at each week from 39 to 44 gestational weeks. The central idea is that, for all cortical surfaces, we first group-wisely register them into the common space to ensure the unbiasedness. Then, rather than simply averaging over the co-registered cortical surfaces, which generally leads to over-smoothed cortical folding patterns, we adopt a spherical patch-based sparse representation using an augmented dictionary to overcome the noises and potential registration errors. Through the group-wise sparsity constraint, we obtain consistent geometric cortical folding attributes on the atlases. Our atlases preserve the sharp cortical folding patterns, thus leading to better registration accuracy when aligning new subjects onto the atlases.

Group-based sparse representation for image compressive sensing reconstruction with non-convex regularization

Patch-based sparse representation modeling has shown great potential in image compressive sensing (CS) reconstruction. However, this model usually suffers from some limits, such as dictionary learning with great computational complexity, neglecting the relationship among similar patches. In this paper, a group-based sparse representation method with non-convex regularization (GSR-NCR) for image CS reconstruction is proposed. In GSR-NCR, the local sparsity and nonlocal self-similarity of images is simultaneously considered in a unified framework. Different from the previous methods based on sparsity-promoting convex regularization, we extend the non-convex weighted Lp (0 < p < 1) penalty function on group sparse coefficients of the data matrix, rather than conventional L1-based regularization. To reduce the computational complexity, instead of learning the dictionary with a high computational complexity from natural images, we learn the principle component analysis (PCA) based dictionary for each group. Moreover, to make the proposed scheme tractable and robust, we have developed an efficient iterative shrinkage/thresholding algorithm to solve the non-convex optimization problem. Experimental results demonstrate that the proposed method outperforms many state-of-the-art techniques for image CS reconstruction.

A note on patch-based low-rank minimization for fast image denoising

Patch-based low-rank minimization for image processing attracts much attention in recent years. The minimization of the matrix rank coupled with the Frobenius norm data fidelity can be solved by the hard thresholding filter with principle component analysis (PCA) or singular value decomposition (SVD). Based on this idea, we propose a patch-based low-rank minimization method for image denoising. The main denoising process is stated in three equivalent way: PCA, SVD and low-rank minimization. Compared to recent patch-based sparse representation methods, experiments demonstrate that the proposed method is rather rapid, and it is effective for a variety of natural grayscale images and color images, especially for texture parts in images. Further improvements of this method are also given. In addition, due to the simplicity of this method, we could provide an explanation of the choice of the threshold parameter, estimation of PSNR values, and give other insights into this method.

Patch-based sparse reconstruction for electrical impedance tomography

PurposeElectrical impedance tomography (EIT) is a technique for reconstructing the conductivity distribution by injecting currents at the boundary of a subject and measuring the resulting changes in voltage. Image reconstruction for EIT is a nonlinear problem. A generalized inverse operator is usually ill-posed and ill-conditioned. Therefore, the solutions for EIT are not unique and highly sensitive to the measurement noise.Design/methodology/approachThis paper develops a novel image reconstruction algorithm for EIT based on patch-based sparse representation. The sparsifying dictionary optimization and image reconstruction are performed alternately. Two patch-based sparsity, namely, square-patch sparsity and column-patch sparsity, are discussed and compared with the global sparsity.FindingsBoth simulation and experimental results indicate that the patch based sparsity method can improve the quality of image reconstruction and tolerate a relatively high level of noise in the measured voltages.Originality/valueEIT image is reconstructed based on patch-based sparse representation. Square-patch sparsity and column-patch sparsity are proposed and compared. Sparse dictionary optimization and image reconstruction are performed alternately. The new method tolerates a relatively high level of noise in measured voltages.

Robust patch-based sparse representation for hyperspectral image classification

Sparse representation classification (SRC) has been successfully applied into hyperspectral image (HSI). A test sample (pixel) can be linearly represented by a few training samples of the training set. The class label of the test sample is then decided by the reconstruction residuals. To incorporate the spatial information to improve the classification performance, a patch matrix, which includes a spatial neighborhood set, is used to replace the original pixel. Generally, the objective function of the reconstruction residuals is represented as Frobenius-norm, which actually treats the elements in the reconstruction residuals in the same way. However, when a patch locates in the image edge, the samples in the patch may belong to different classes. Frobenius-norm is not suitable to compute the reconstruction residuals. In this paper, we propose a robust patch-based sparse representation classification (RPSRC) based on [Formula: see text]-norm. An iteration algorithm is given to compute RPSRC efficiently. Extensive experimental results on two real-life HSI datasets demonstrate the effectiveness of RPSRC.

Block compressive sensing of image and video with nonlocal Lagrangian multiplier and patch-based sparse representation

Although block compressive sensing (BCS) makes it tractable to sense large-sized images and video, its recovery performance has yet to be significantly improved because its recovered images or video usually suffer from blurred edges, loss of details, and high-frequency oscillatory artifacts, especially at a low subrate. This paper addresses these problems by designing a modified total variation technique that employs multi-block gradient processing, a denoised Lagrangian multiplier, and patch-based sparse representation. In the case of video, the proposed recovery method is able to exploit both spatial and temporal similarities. Simulation results confirm the improved performance of the proposed method for compressive sensing of images and video in terms of both objective and subjective qualities.

4D Infant Cortical Surface Atlas Construction using Spherical Patch-based Sparse Representation.

The 4D infant cortical surface atlas with densely sampled time points is highly needed for neuroimaging analysis of early brain development. In this paper, we build the 4D infant cortical surface atlas firstly covering 6 postnatal years with 11 time points (i.e., 1, 3, 6, 9, 12, 18, 24, 36, 48, 60, and 72 months), based on 339 longitudinal MRI scans from 50 healthy infants. To build the 4D cortical surface atlas, first, we adopt a two-stage groupwise surface registration strategy to ensure both longitudinal consistency and unbiasedness. Second, instead of simply averaging over the co-registered surfaces, a spherical patch-based sparse representation is developed to overcome possible surface registration errors across different subjects. The central idea is that, for each local spherical patch in the atlas space, we build a dictionary, which includes the samples of current local patches and their spatially-neighboring patches of all co-registered surfaces, and then the current local patch in the atlas is sparsely represented using the built dictionary. Compared to the atlas built with the conventional methods, the 4D infant cortical surface atlas constructed by our method preserves more details of cortical folding patterns, thus leading to boosted accuracy in registration of new infant cortical surfaces.

Accelerated reconstruction of electrical impedance tomography images via patch based sparse representation.

Electrical impedance tomography (EIT) reconstruction is a nonlinear and ill-posed problem. Exact reconstruction of an EIT image inverts a high dimensional mathematical model to calculate the conductivity field, which causes significant problems regarding that the computational complexity will reduce the achievable frame rate, which is considered as a major advantage of EIT imaging. The single-step method, state estimation method, and projection method were always used to accelerate reconstruction process. The basic principle of these methods is to reduce computational complexity. However, maintaining high resolution in space together with not much cost is still challenging, especially for complex conductivity distribution. This study proposes an idea to accelerate image reconstruction of EIT based on compressive sensing (CS) theory, namely, CSEIT method. The novel CSEIT method reduces the sampling rate through minimizing redundancy in measurements, so that detailed information of reconstruction is not lost. In order to obtain sparse solution, which is the prior condition of signal recovery required by CS theory, a novel image reconstruction algorithm based on patch-based sparse representation is proposed. By applying the new framework of CSEIT, the data acquisition time, or the sampling rate, is reduced by more than two times, while the accuracy of reconstruction is significantly improved.

Accelerated Split Bregman Method for Image Compressive Sensing Recovery under Sparse Representation

Compared with traditional patch-based sparse representation, recent studies have concluded that group-based sparse representation (GSR) can simultaneously enforce the intrinsic local sparsity and nonlocal self-similarity of images within a unified framework. This article investigates an accelerated split Bregman method (SBM) that is based on GSR which exploits image compressive sensing (CS). The computational efficiency of accelerated SBM for the measurement matrix of a partial Fourier matrix can be further improved by the introduction of a fast Fourier transform (FFT) to derive the enhanced algorithm. In addition, we provide convergence analysis for the proposed method. Experimental results demonstrate that accelerated SBM is potentially faster than some existing image CS reconstruction methods.