The quintessential goal of sensor array signal processing is the estimation parameters by fusing temporal and spatial information, captured via sampling a wavefield with set judiciously placed antenna sensors. assumed to be generated finite number emitters, contains information about characterizing emitters. A review area given. focus on parameter methods, many relevant problems are only briefly mentioned. We emphasize relatively more recent subspace-based methods in relation beamforming....

A simple construction of an orthonormal basis starting with a so-called mother wavelet, together efficient implementation gained the wavelet decomposition easy acceptance and generated great research interest in its applications. An may not, however, always be suitable representation signal, particularly when time (or space) invariance is required property. The conventional way around this problem to use redundant decomposition. We address time-invariance for transforms propose extension...

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> It is well known that the wavelet transform provides a very effective framework for analysis of multiscale edges. In this paper, we propose novel approach based on shearlet transform: directional with greater ability to localize distributed discontinuities such as Indeed, unlike traditional wavelets, shearlets are theoretically optimal in representing images edges and, particular, have fully...

We propose a best basis algorithm for signal enhancement in white Gaussian noise. The search is performed families of orthonormal bases constructed with wavelet packets or local cosine bases. base our the "best" on criterion minimal reconstruction error underlying signal. This approach intuitively appealing, because enhanced estimated has an associated measure performance, namely, resulting mean-square error. Previous approaches this framework have focused obtaining most "compact"...

Deep dictionary learning seeks multiple dictionaries at different image scales to capture complementary coherent characteristics. We propose a method for hierarchy of synthesis with an classification goal. The and parameters are trained by objective, the sparse features extracted reducing reconstruction loss in each layer. objectives some sense regularize problem inject source signal information features. performance proposed hierarchical increases adding more layers, which consequently...

Nonlinear filtering techniques based on the theory of robust estimation are introduced. Some deterministic and asymptotic properties derived. The proposed denoising methods optimal over Huber /spl epsi/-contaminated normal neighborhood highly resistant to outliers. Experimental results showing a much improved performance filters in presence Gaussian heavy-tailed noise analyzed illustrated.

Entropy-based divergence measures have shown promising results in many areas of engineering and image processing. We define a new generalized measure, namely, the Jensen-Renyi (1996, 1976) divergence. Some properties such as convexity its upper bound are derived. Based on divergence, we propose approach to problem registration. appealing advantages registration by illustrated, connections mutual information-based techniques analyzed. As key focus this paper, apply for inverse synthetic...

We introduce a family of first-order multidimensional ordinary differential equations (ODEs) with discontinuous right-hand sides and demonstrate their applicability in image processing. An equation belonging to this is an inverse diffusion everywhere except at local extrema, where some stabilization introduced. For reason, we call these "stabilized equations" (SIDEs). Existence uniqueness solutions, as well stability, are proven for SIDEs. A SIDE one spatial dimension may be interpreted...

Pixel level image fusion refers to the processing and synergistic combination of information gathered by various imaging sources provide a better understanding scene. We formulate as an optimization problem propose inf

The authors carry out a performance analysis of two eigenstructure-based direction-of-arrival estimation algorithms, using series expansion projection operators (or projectors) on the signal and noise subspaces. In interest algebraic simplicity, an operator formalism is utilized. A perturbation performed projectors, results which are used to determine effect estimated parameters. approach makes it possible any chosen order projectors by original recurrence formula developed for higher-order...

We present efficient multiscale approaches to the segmentation of natural clutter, specifically grass and forest, enhancement anomalies in synthetic aperture radar (SAR) imagery. The methods we propose exploit coherent nature SAR sensors. In particular, they take advantage characteristic statistical differences imagery different terrain types, as a function scale, due speckle. employ class stochastic processes that provide powerful framework for describing random fields evolve scale. build...

Data heterogeneity can pose a great challenge to process and systematically fuse low-level data from different modalities with no recourse heuristics manual adjustments refinements. In this paper, new methodology is introduced for the fusion of measured detecting predicting weather-driven natural hazards. The proposed research introduces robust theoretical algorithmic framework heterogeneous in near real time. We establish flexible information-based target optimality criterion choice, which...

We propose a new approach to Generative Adversarial Networks (GANs) achieve an improved performance with additional robustness its so-called and well-recognized mode collapse. first proceed by mapping the desired data onto frame-based space for sparse representation lift any limitation of small support features prior learning structure. To that end, we start dividing image into multiple patches modifying role generative network from producing entire image, at once, creating vector each...

A discriminative structured analysis dictionary is proposed for the classification task. structure of union subspaces (UoS) integrated into conventional learning to enhance capability discrimination. simple classifier also simultaneously included formulated function ensure a more complete consistent classification. The solution algorithm efficiently obtained by linearized alternating direction method multipliers. Moreover, distributed presented address large-scale datasets. It can...

Approaches to wavelet-based denoising (or signal enhancement) have generally relied on the assumption of normally distributed perturbations. To relax this assumption, which is often violated in practice, we derive a robust wavelet thresholding technique based minimax description length (MMDL) principle. We first determine least favorable distribution /spl epsiv/-contaminated normal family as member that maximizes entropy. show distribution, and best estimate upon it, namely...

Recognition of images and shapes has long been the central theme computer vision. Its importance is increasing rapidly in field graphics multimedia communication because it difficult to process information efficiently without its recognition. In this paper, we propose a new approach for object matching based on global geodesic measure. The key idea behind our methodology represent an by probabilistic shape descriptor that measures distance between two arbitrary points surface object....

Wavelet packets and local trigonometric bases provide an efficient framework fast algorithms to obtain a "best basis" or representation" of deterministic signals. Applying these techniques stochastic processes may, however, lead variable results. We revisit this problem introduce prior model on the underlying signal in noise account for contaminating as well. thus develop Bayesian-based approach best basis problem, while preserving classical tree search efficiency.

A maximum a posteriori (MAP) estimator using Markov or entropy random field model for prior distribution may be viewed as minimizer of variational problem.Using notions from robust statistics, filter referred to Huber gradient descent flow is proposed. It result optimizing functional subject some noise constraints and takes hybrid form total variation diffusion large magnitudes linear small magnitudes. Using the gained insight, further extension, we propose an information-theoretic which...