A5048650003- Image and Signal Denoising Methods
- Advanced Image Processing Techniques
- Advanced Vision and Imaging
- Video Coding and Compression Technologies
- Sparse and Compressive Sensing Techniques
- Advanced Data Compression Techniques
- Image Processing Techniques and Applications
- Advanced Image Fusion Techniques
- Medical Image Segmentation Techniques
- Advanced Image and Video Retrieval Techniques
- Neural Networks and Applications
- Advanced Wireless Network Optimization
- Blind Source Separation Techniques
- Photoacoustic and Ultrasonic Imaging
- Video Analysis and Summarization
- Speech and Audio Processing
- Medical Imaging Techniques and Applications
- Image and Video Quality Assessment
- Optical measurement and interference techniques
- Advanced MRI Techniques and Applications
- Cooperative Communication and Network Coding
- Music and Audio Processing
- Image Enhancement Techniques
- Advanced X-ray and CT Imaging
- Image Retrieval and Classification Techniques
Northwestern University
2016-2025
Intel (United States)
2021-2024
Science North
2012-2024
Northwestern University
2009-2023
University of Arizona
2023
Universidad de Cádiz
2023
Argonne National Laboratory
2010-2021
McCormick (United States)
1992-2020
Northwest University
2019
Universidad de Granada
1999-2017
The article introduces digital image restoration to the reader who is just beginning in this field, and provides a review analysis for may already be well-versed restoration. perspective on topic one that comes primarily from work done field of signal processing. Thus, many techniques works cited relate classical processing approaches estimation theory, filtering, numerical analysis. In particular, emphasis placed algorithms grow out an area known as "regularized least squares" methods. It...
In this paper, we model the components of compressive sensing (CS) problem, i.e., signal acquisition process, unknown coefficients and parameters for noise using Bayesian framework. We utilize a hierarchical form Laplace prior to sparsity signal. describe relationship among number priors proposed in literature, show advantages including its high degree sparsity. Moreover, that some existing models are special cases model. Using our model, develop constructive (greedy) algorithm designed fast...
A hybrid multidimensional image segmentation algorithm is proposed, which combines edge and region-based techniques through the morphological of watersheds. An edge-preserving statistical noise reduction approach used as a preprocessing stage in order to compute an accurate estimate gradient. Then, initial partitioning into primitive regions produced by applying watershed transform on gradient magnitude. This input computationally efficient hierarchical (bottom-up) region merging process...
Convolutional neural networks (CNN) are a special type of deep (DNN). They have so far been successfully applied to image super-resolution (SR) as well other restoration tasks. In this paper, we consider the problem video super-resolution. We propose CNN that is trained on both spatial and temporal dimensions videos enhance their resolution. Consecutive frames motion compensated used input provides super-resolved output. investigate different options combining within one architecture. While...
Traditionally, analytical methods have been used to solve imaging problems such as image restoration, inpainting, and superresolution (SR). In recent years, the fields of machine deep learning gained a lot momentum in solving problems, often surpassing performance provided by approaches. Unlike for which problem is explicitly defined domain-knowledge carefully engineered into solution, neural networks (DNNs) do not benefit from prior knowledge instead make use large data sets learn unknown...
We review error resilience techniques for real-time video transport over unreliable networks. Topics covered include an introduction to today's protocol and network environments their characteristics, encoder tools, decoder concealment techniques, as well that require cooperation between encoder, decoder, the network. provide a of general principles these specific implementations adopted by H.263 MPEG-4 coding standards. The majority article is devoted developed block-based hybrid coders...
The application of regularization to ill-conditioned problems necessitates the choice a parameter which trades fidelity data with smoothness solution. value depends on variance noise in data. problem choosing and estimating image restoration is examined. An error analysis based an objective mean-square-error (MSE) criterion used motivate regularization. Two approaches for are proposed. proposed existing methods compared their relationship linear minimum-mean-square-error filtering...
Video object segmentation targets segmenting a specific throughout video sequence when given only an annotated first frame. Recent deep learning based approaches find it effective to fine-tune general-purpose model on the frame using hundreds of iterations gradient descent. Despite high accuracy that these methods achieve, fine-tuning process is inefficient and fails meet requirements real world applications. We propose novel approach uses single forward pass adapt appearance object....
Recovery of low-rank matrices has recently seen significant activity in many areas science and engineering, motivated by recent theoretical results for exact reconstruction guarantees interesting practical applications. In this paper, we present novel recovery algorithms estimating matrix completion robust principal component analysis based on sparse Bayesian learning (SBL) principles. Starting from a factorization formulation enforcing the constraint estimates as sparsity constraint,...
With the first direct detection of gravitational waves, advanced laser interferometer gravitational-wave observatory (LIGO) has initiated a new field astronomy by providing an alternative means sensing universe. The extreme sensitivity required to make such detections is achieved through exquisite isolation all sensitive components LIGO from non-gravitational-wave disturbances. Nonetheless, still susceptible variety instrumental and environmental sources noise that contaminate data. Of...
Abstract There exist a number of satellites on different earth observation platforms, which provide multispectral images together with panchromatic image, that is, an image containing reflectance data representative wide range bands and wavelengths. Pansharpening is pixel-level fusion technique used to increase the spatial resolution while simultaneously preserving its spectral information. In this paper, we review pan-sharpening methods proposed in literature giving clear classification...
In this paper, we address the super resolution (SR) problem from a set of degraded low (LR) images to obtain high (HR) image. Accurate estimation sub-pixel motion between LR significantly affects performance reconstructed HR propose novel methods where image and parameters are estimated simultaneously. Utilizing Bayesian formulation, model unknown image, acquisition process, in stochastic sense. Employing variational analysis, develop two algorithms which jointly estimate distributions all...
Video super-resolution (VSR) has become one of the most critical problems in video processing. In deep learning literature, recent works have shown benefits using adversarial-based and perceptual losses to improve performance on various image restoration tasks; however, these yet be applied for super-resolution. this paper, we propose a generative adversarial network (GAN)-based formulation VSR. We introduce new generator optimized VSR problem, named VSRResNet, along with discriminator...
Background There are characteristic findings of coronavirus disease 2019 (COVID-19) on chest images. An artificial intelligence (AI) algorithm to detect COVID-19 radiographs might be useful for triage or infection control within a hospital setting, but prior reports have been limited by small data sets, poor quality, both. Purpose To present DeepCOVID-XR, deep learning AI radiographs, that was trained and tested large clinical set. Materials Methods DeepCOVID-XR is an ensemble convolutional...
Respiratory diseases constitute one of the leading causes death worldwide and directly affect patient's quality life. Early diagnosis patient monitoring, which conventionally include lung auscultation, are essential for efficient management respiratory diseases. Manual sound interpretation is a subjective time-consuming process that requires high medical expertise. The capabilities deep learning offers could be exploited in order robust classification models can designed. In this paper, we...
Most massive stars are members of a binary or higher-order stellar systems, where the presence companion can decisively alter their evolution via interactions. Interacting binaries also important astrophysical laboratories for study compact objects. Binary population synthesis studies have been used extensively over last two decades to interpret observations compact-object and decipher physical processes that lead formation. Here, we present POSYDON, novel, code incorporates full...
The reconstruction of images from incomplete block discrete cosine transform (BDCT) data is examined. problem formulated as one regularized image recovery. According to this formulation, the in decoder reconstructed by using not only transmitted but also prior knowledge about smoothness original image, which complements data. Two methods are proposed for solving recovery problem. first based on theory projections onto convex sets (POCS) while second constrained least squares (CLS) approach....
At the present time, block-transform coding is probably most popular approach for image compression. For this approach, compressed images are decoded using only transmitted transform data. We formulate decoding as an recovery problem. According to reconstructed not data but, in addition, prior knowledge that before compression do display between-block discontinuities. A spatially adaptive algorithm proposed based on theory of projections onto convex sets. Apart from constraint set, uses...
The development of the algorithm is based on a set theoretic approach to regularization. Deterministic and/or statistical information about undistorted image and noise are directly incorporated into iterative procedure. restored center an ellipsoid bounding intersection two ellipsoids. proposed algorithm, which has constrained least squares as special case, extended adaptive restoration algorithm. spatial adaptivity introduced incorporate properties human visual system. Convergence...
This tutorial paper discusses the use of successive-approximation-based iterative restoration algorithms for removal linear blurs and noise from images. Iterative are particularly attractive this application because they allow incorporation prior knowledge about class feasible solutions, can be used to remove nonstationary blurs, fairly robust with respect errors in approximation blurring operator. Regularization is introduced as a means preventing excessive magnification that typically...
A modified Hopfield neural network model for regularized image restoration is presented. The proposed allows negative autoconnections each neuron. set of algorithms using the presented, with various updating modes: sequential updates; n-simultaneous and partially asynchronous updates. algorithm shown to converge a local minimum energy function after finite number iterations. Since an which updates all n neurons simultaneously not guaranteed converge, called greedy algorithm. Although...
The performance of an automatic facial expression recognition system can be significantly improved by modeling the reliability different streams information utilizing multistream hidden Markov models (HMMs). In this paper, we present HMM and analyze its performance. proposed utilizes animation parameters (FAPs), supported MPEG-4 standard, as features for classification. Specifically, FAPs describing movement outer-lip contours eyebrows are used observations. Experiments first performed...