Separate 3D Research Articles

Dynamic Energy, Performance, and Accuracy Optimization and Management Using Automatically Generated Constraints for Separable 2D FIR Filtering for Digital Video Processing

There is strong interest in the development of dynamically reconfigurable systems that can meet real-time constraints on energy, performance, and accuracy. The generation of real-time constraints will significantly expand the applicability of dynamically reconfigurable systems to new domains, such as digital video processing. We develop a dynamically reconfigurable 2D FIR filtering system that can meet real-time constraints in energy, performance, and accuracy (EPA). The real-time constraints are automatically generated based on user input, image types associated with video communications, and video content. We first generate a set of Pareto-optimal realizations, described by their EPA values and associated 2D FIR hardware description bitstreams. Dynamic management is then achieved by selecting Pareto-optimal realizations that meet the automatically generated time-varying EPA constraints. We validate our approach using three different 2D Gaussian filters. Filter realizations are evaluated in terms of the required energy per frame, accuracy of the resulting image, and performance in frames per second. We demonstrate dynamic EPA management by applying a Difference of Gaussians (DOG) filter to standard video sequences. For video frame sizes that are equal to or larger than the VGA resolution, compared to a static implementation, our dynamic system provides significant reduction in the total energy consumption (>30%).

Wavelet-Based 3D Reconstruction of Microcalcification Clusters from Two Mammographic Views: New Evidence That Fractal Tumors Are Malignant and Euclidean Tumors Are Benign

The 2D Wavelet-Transform Modulus Maxima (WTMM) method was used to detect microcalcifications (MC) in human breast tissue seen in mammograms and to characterize the fractal geometry of benign and malignant MC clusters. This was done in the context of a preliminary analysis of a small dataset, via a novel way to partition the wavelet-transform space-scale skeleton. For the first time, the estimated 3D fractal structure of a breast lesion was inferred by pairing the information from two separate 2D projected mammographic views of the same breast, i.e. the cranial-caudal (CC) and mediolateral-oblique (MLO) views. As a novelty, we define the “CC-MLO fractal dimension plot”, where a “fractal zone” and “Euclidean zones” (non-fractal) are defined. 118 images (59 cases, 25 malignant and 34 benign) obtained from a digital databank of mammograms with known radiologist diagnostics were analyzed to determine which cases would be plotted in the fractal zone and which cases would fall in the Euclidean zones. 92% of malignant breast lesions studied (23 out of 25 cases) were in the fractal zone while 88% of the benign lesions were in the Euclidean zones (30 out of 34 cases). Furthermore, a Bayesian statistical analysis shows that, with 95% credibility, the probability that fractal breast lesions are malignant is between 74% and 98%. Alternatively, with 95% credibility, the probability that Euclidean breast lesions are benign is between 76% and 96%. These results support the notion that the fractal structure of malignant tumors is more likely to be associated with an invasive behavior into the surrounding tissue compared to the less invasive, Euclidean structure of benign tumors. Finally, based on indirect 3D reconstructions from the 2D views, we conjecture that all breast tumors considered in this study, benign and malignant, fractal or Euclidean, restrict their growth to 2-dimensional manifolds within the breast tissue.

A Self-Reconfigurable Platform for the Implementation of 2D Filterbanks with Real and Complex-Valued Inputs, Outputs, and Filter Coefficients

We introduce a dynamically reconfigurable 2D filterbank that supports both real and complex-valued inputs, outputs, and filter coefficients. This general purpose filterbank allows for the efficient implementation of 2D filterbanks based on separable 2D FIR filters that support all possible combinations of input and output signals. The system relies on the use of dynamic reconfiguration of real/complex one-dimensional filters to minimize the required hardware resources. The system is demonstrated using an equiripple and a Gabor filterbank and the results using both real and complex-valued input images. We summarize the performance of the system in terms of the required processing times, energy, and accuracy.

ATHENA 3D: A finite element code for ultrasonic wave propagation

The understanding of wave propagation phenomena requires use of robust numerical models. 3D finite element (FE) models are generally prohibitively time consuming. However, advances in computing processor speed and memory allow them to be more and more competitive. In this context, EDF R&D developed the 3D version of the well-validated FE code ATHENA2D. The code is dedicated to the simulation of wave propagation in all kinds of elastic media and in particular, heterogeneous and anisotropic materials like welds. It is based on solving elastodynamic equations in the calculation zone expressed in terms of stress and particle velocities. The particularity of the code relies on the fact that the discretization of the calculation domain uses a Cartesian regular 3D mesh while the defect of complex geometry can be described using a separate (2D) mesh using the fictitious domains method. This allows combining the rapidity of regular meshes computation with the capability of modelling arbitrary shaped defects. Furthermore, the calculation domain is discretized with a quasi-explicit time evolution scheme. Thereby only local linear systems of small size have to be solved. The final step to reduce the computation time relies on the fact that ATHENA3D has been parallelized and adapted to the use of HPC resources. In this paper, the validation of the 3D FE model is discussed. A cross-validation of ATHENA 3D and CIVA is proposed for several inspection configurations. The performances in terms of calculation time are also presented in the cases of both local computer and computation cluster use.

Image quality assessment based on matching pursuit

The objective assessment of image quality is an essential part of many visual processing systems. The challenge lies in evaluating the image quality consistently under subjective perceptions. In this paper, we propose a novel image quality metric based on the matching pursuit algorithm. Under the principle of structural information distortion, we assume that various structure data contributes differently to the single image quality score. Specifically, we decompose the reference image using matching pursuit with a separable 2D Gabor dictionary, thus obtaining structural information, and develop a characterization for this information and its importance. We then discuss the relationship between the structural distortion intensity and the subjective quality measurement on the energy scale. In experiments, we compare the performance of our algorithm with both subjective ratings and state-of-the-art objective methods on image datasets with multi-type distortions. The experimental results validate our proposed method.

Variational Image Registration Using Inhomogeneous Regularization

We present a generalization of the convolution-based variational image registration approach, in which different regularizers can be implemented by conveniently exchanging the convolution kernel, even if it is nonseparable or nonstationary. Nonseparable kernels pose a challenge because they cannot be efficiently implemented by separate 1D convolutions. We propose to use a low-rank tensor decomposition to efficiently approximate nonseparable convolution. Nonstationary kernels pose an even greater challenge because the convolution kernel depends on, and needs to be evaluated for, every point in the image. We propose to pre-compute the local kernels and efficiently store them in memory using the Tucker tensor decomposition model. In our experiments we use the nonseparable exponential kernel and a nonstationary landmark kernel. The exponential kernel replicates desirable properties of elastic image registration, while the landmark kernel incorporates local prior knowledge about corresponding points in the images. We examine the trade-off between the computational resources needed and the approximation accuracy of the tensor decomposition methods. Furthermore, we obtain very smooth displacement fields even in the presence of large landmark displacements.

Reduction of Lifting Steps of Non Separable 2D Quadruple Lifting DWT Compatible with Separable 2D DWT

Reduction of Lifting Steps of Non Separable 2D Quadruple Lifting DWT Compatible with Separable 2D DWT

A theoretical deduction of the shape and size of nanocarbons suitable for hydrogen storage

We evaluated the adsorption energy of a hydrogen molecule in nanocarbons consisting of graphene sheets. The nanocarbon shapes were a pair of disks with separation 2d, a cylinder with radius d, and a truncated sphere with radius d. We obtained the adsorption energy in the form of a 10–4 Lennard–Jones function with respect to 1/d. The values of the potential depth (D) and equilibrium distance (de), respectively, were 94 meV and 2.89 A for the disk pair, 158 meV and 3.14 A for the cylinder, and 203 meV and 3.37 A for the sphere. When d=de, the adsorption energy of the disk pair (cylinder) became deeper than −0.9D, and it approached −D when the radius (length) increased to more than twice its separation (radius). The adsorption energy of the sphere was increased from −D to −0.5D when the radius of the opening increased from 0 to de. These results suggest that porous carbon materials can increase the adsorption energy by up to ∼200 meV if the carbon atoms are arranged on a spherical-like surface with ∼7 A separation. This may lead to practical hydrogen storage for fuel cells.

Thermal Analysis of AlGaN/GaN HEMTs Using Angular Fourier-Series Expansion

Thermal analysis of planar and near-square semiconductor device chips employing angular Fourier-series (AFS) expansion is presented for the first time. The determination of the device peak temperature using AFS requires only a single two-dimensional computation, while full three-dimensional temperature distribution can be obtained, if desired, by successively adding higher-order Fourier terms, each of which requires a separate 2D computation. The AFS method is used to compare the heat spreading characteristics of AlGaN/GaN high-electron-mobility transistors (HEMTs) fabricated on silicon, silicon carbide, and synthetic diamond. We show that AlGaN/GaN HEMTs built using GaN/diamond technology can offer better than half the thermal resistance of GaN/SiC HEMTs under worst-case cooling conditions. Furthermore, we show that, if left unmanaged, an inherent and non-negligible thermal boundary resistance due to the integration of semiconductor epilayers with non-native substrates will dampen the benefits of highly conductive substrates such as SiC and diamond.

Fast Gabor texture feature extraction with separable filters using GPU

Gabor wavelet transform is one of the most effective texture feature extraction techniques and has resulted in many successful practical applications. However, real-time applications cannot benefit from this technique because of the high computational cost arising from the large number of small-sized convolutions which require over 10 min to process an image of 256 × 256 pixels on a dual core CPU. As the computation in Gabor filtering is parallelizable, it is possible and beneficial to accelerate the feature extraction process using GPU. Conventionally, this can be achieved simply by accelerating the 2D convolution directly, or by expediting the CPU-efficient FFT-based 2D convolution. Indeed, the latter approach, when implemented with small-sized Gabor filters, cannot fully exploit the parallel computation power of GPU due to the architecture of graphics hardware. This paper proposes a novel approach tailored for GPU acceleration of the texture feature extraction algorithm by using separable 1D Gabor filters to approximate the non-separable Gabor filter kernels. Experimental results show that the approach improves the timing performance significantly with minimal error introduced. The method is specifically designed and optimized for computing unified device architecture and is able to achieve a speed of 16 fps on modest graphics hardware for an image of 2562 pixels and a filter kernel of 322 pixels. It is potentially applicable for real-time applications in areas such as motion tracking and medical image analysis.

TU‐A‐116‐02: A Novel Algorithm for Highly Efficient Polyenergetic DRR Calculation

Purpose: It is desirable to compute polyenergetic digitally reconstructed radiographs (DRRs) in many radiotherapy tasks, especially when investigating beam hardening effects in CT/CBCT reconstruction and studying 2D‐3D registration. The standard polyenergetic DRR calculation performs ray‐tracing for each energy bin. The computation time scales with the number of bins in the spectrum and the ray tracing operation for each bin dominates the computation time. This work develops a novel algorithm to enhance the efficiency of polyenergetic DRR computation. Methods: The inputs of our algorithm are 3D density and material arrays along with the energy spectrum. We compute a modified projection by ray tracing through the density data and saving the results from each material into separate 2D arrays. For each energy bin, the product of each density projection and its corresponding energy dependent mass attenuation coefficient is summed over all materials. The exponential of the results from each energy bin are weighted and summed to create a final polyenergetic DRR. Our new algorithm is mathematically equivalent to the standard approach. We use our algorithm to simulate a 512×512 polyenergetic DRR of a 512×512×182 phantom comprised of 3 materials (air, tissue, bone). A realistic 120kVp energy spectrum with 114 bins was used. The computation times and results of the new algorithm are compared to those of the standard algorithm. Results: We successfully improved the efficiency of the polyenergetic DRR calculation by a factor of 63.85 in our testing case. The average error between the results from our algorithm and that from the standard algorithm is on the order of machine precision. Conclusion: We have substantially improved the efficiency of the standard polyenergetic DRR algorithm without sacrificing accuracy. This new approach enables fast and accurate DRR calculations that will facilitate many tasks in radiotherapy.

Tomographic reconstruction of the full 4D transverse phase space

We present a method to measure and reconstruct the full 4D transverse phase space of a particle beam. The method assumes that it is possible to freely and independently rotate the separate 2D transverse phase spaces in the horizontal and vertical directions. Using images of the beam that would be captured on a screen, we develop a mathematical procedure that will reconstruct the full 4D particle distribution. We simulate this procedure for a hypothetical distribution and show that the reconstruction agrees with the actual distribution. Finally, we demonstrate for the practical case of a two-quadrupole setup that it is indeed possible to adjust the quadrupole strengths so that the separate 2D transverse phase spaces can be rotated independently. However, in this setup, the rotation angles are restricted to smaller ranges. Even so, the reconstructed results reproduce the actual distribution clearly.

3D cine phase-contrast MRI at 3T in intracranial aneurysms compared with patient-specific computational fluid dynamics.

CFD has been proved valuable for simulating blood flow in intracranial aneurysms, which may add to better rupture risk assessment. However, CFD has drawbacks such as the sensitivity to assumptions needed for the model, which may hinder its clinical implementation. 3D PC-MR imaging is a technique that enables measurements of blood flow. The purpose of this study was to compare flow patterns on the basis of 3D PC-MR imaging with CFD estimates. 3D PC-MR imaging was performed in 8 intracranial aneurysms. Two sets of patient-specific inflow boundaries for CFD were obtained from a separate 2D PC-MR imaging sequence (2D CFD) and from the 3D PC-MR imaging (3D CFD) data. 3D PC-MR imaging and CFD were compared by calculation of the differences between velocity vector magnitudes and angles. Differences in flow patterns expressed as the presence and strengths of vortices were determined by calculation of singular flow energy. In systole, flow features such as vortex patterns were similar. In diastole, 3D PC-MR imaging measurements appeared inconsistent due to low velocity-to-noise ratios. The relative difference in velocity magnitude was 67.6 ± 51.4% and 27.1 ± 24.9% in systole and 33.7 ± 21.5% and 17.7 ± 10.2% in diastole for 2D CFD and 3D CFD, respectively. For singular energy, this was reduced to 15.5 ± 13.9% at systole and 19.4 ± 17.6% at diastole (2D CFD). In systole, good agreement between 3D PC-MR imaging and CFD on flow-pattern visualization and singular-energy calculation was found. In diastole, flow patterns of 3D PC-MR imaging differed from those obtained from CFD due to low velocity-to-noise ratios.

Two dimensional bias correction of temperature and precipitation copulas in climate models

In common climate model bias‐correction procedures, temperature and precipitation are corrected separately, thereby degrading the dynamical link represented within the model. We propose a methodology that advances the state‐of‐the‐art by correcting not just the 1D intensity distributions separately but the full two‐dimensional statistical distribution. To assess the effectiveness of the proposed method, it is applied to the REMO regional climate model output using point measurements of hourly temperature and precipitation from 6 weather stations over Germany as observations. A standard cross‐validation is performed by dividing the data into two nonoverlapping 15 year periods. Results show that the methodology effectively improves the temperature‐precipitation copula in the validation period, unlike separate 1D temperature and precipitation corrections which, by construction, leave the copula unchanged. An unexpected result is that a relatively small number (<5) of temperature bins are required to achieve significant improvements in the copula. Results are similar for all stations.

Quantum 3D spin-glass system on the scales of space-time periods of external electromagnetic fields

A dielectric medium consisting of rigidly polarized molecules has been treated as a quantum 3D disordered spin system. It is shown that using Birkhoff’s ergodic hypothesis the initial 3D disordered spin problem on scales of space-time periods of external field is reduced to two conditionally separable 1D problems. The first problem describes a 1D disordered N-particle quantum system with relaxation in random environment while the second one describes statistical properties of ensemble of disordered 1D steric spin chains of certain length. Basing on constructions which are developed in both problems, the coefficient of polarizability related to collective orientational effects under the influence of external field was calculated. On the basis of these investigations the equation of Clausius-Mossotti (CM) has been generalized as well as the equation for permittivity. It is shown that under the influence of weak standing electromagnetic fields in the equation of CM arising of catastrophe is possible, that can substantially change behavior of permittivity in the X-ray region on the macroscopic scale of space.

Robust geophysical integration through structure-coupled joint inversion and multispectral fusion of seismic reflection, magnetotelluric, magnetic, and gravity images: Example from Santos Basin, offshore Brazil

We have applied a crossgradient joint inversion and geospectral visualization method to marine seismic reflection, magnetotelluric, gravity, and magnetic data sets acquired along a 162 km profile across a segment of Santos Basin oil province in the continental margin of southeast Brazil. The main exploration targets are the top of the fractured Precambrian crystalline basement and any concealed basement grabens, the overlying presalt and salt/carbonate deposits, and the postsalt cover deposits. The results of joint inversion clearly mapped the various units and are a significant improvement over previous models derived from separate 2D seismic reflection processing and 2D magnetotelluric imaging. Additionally, multispectral fusion of these models resulted in a single image that permits highly constrained geologic interpretations enabling a better understanding of basin architecture. We suggest that joint inversion and image fusion is the way forward for effective geophysical integration.

Compression of 3D Integral Images Using 3D Wavelet Transform

Integral imaging is a technique capable of displaying 3D images with continuous parallax in full natural color. It has been reported by many research groups and is becoming a viable alternative for 3D television. With the development of 3D integral imaging, image compression becomes mandatory for the storage and transmission of 3D integral images. In this paper, the use of the lifting scheme in the application of a 3D Wavelet Transform for the compression of 3D Integral Images is proposed. The method requires the extraction of different viewpoint images from an integral image. The 3D wavelet decomposition is computed by applying three separate 1D transforms along the coordinate axes of the given sequence of Viewpoint Images. The spatial wavelet decompositions on a single viewpoint and on the inter-viewpoint images are performed using the biorthogonal Cohen-Debauchies-Feauveau 9/7 and 5/3 filter banks, respectively. All the resulting wavelet coefficients from application of the 3D wavelet decomposition are arithmetic encoded. Simulations are performed on a set of different grey level 3D Integral Images using a uniform scalar quantizer with deadzone. The results for the average of the four intensity distributions are presented and compared with previous use of 2D DWT and 3D-DCT based schemes. It was found that the algorithm achieves better rate-distortion performance and reconstructs the images with much better image quality at very low bit rates.

Quantitative analysis of vascularization in the finger joints in patients with rheumatoid arthritis using three-dimensional volumetric ultrasonography with power Doppler

This study aimed to compare the intraobserver and interobserver reliability of three-dimensional (3D) and two-dimensional (2D) power Doppler ultrasonography (PDUS) and to assess the relationship between 3D PDUS and clinical parameters in patients with rheumatoid arthritis (RA). Bilateral second/third metacarpophalangeal joints and second/third proximal interphalangeal joints in 33 patients were examined by both 2D and 3D PDUS. Each joint was given a separate 2D PDUS subjective score (range, 0-3) in a standard manner. The 2D PDUS index is the sum of the scores of all eight joints assessed. 3D PDUS voxel signals were quantitatively analyzed by using computerized voxel counts. Intraobserver reliability was high for both examinations (2D PDUS: ICC = 0.957, 95% confidence interval = 0.818-0.999; 3D PDUS: ICC = 0.998, 95% confidence interval = 0.998-1.000). Interobserver reliability was also high (2D PDUS: ICC = 0.993, 95% confidence interval = 0.806-0.988; 3D PDUS: ICC = 0.999, 95% confidence interval = 0.999-1.000). A significant correlation was found between the 2D PDUS index and 3D PDUS voxel count (r = 0.795; p < 0.001). The 3D PDUS voxel count showed significant correlation with 28 joints Disease Activity Score (DAS28)-erythrocyte sedimentation rate (r = 0.448, p < 0.01) and DAS28-C-reactive protein (r = 0.383, p < 0.05). Our study indicates that the measurement of 3D PDUS may be a valuable tool for predicting disease activity.

Image Retrieval using 2D Dual-Tree Discrete Wavelet Transform

large amount of image collections available from a variety of sources has posed increasing technical challenges to computer systems to store/transmit and index/manage the image data to make such collections easily accessible. Here to search and retrieve the expected images from the database we need Content Based Image Retrieval system. This paper proposes a new feature vector based on 2D Dual-tree Discrete Wavelet Transform. One of the advantages of the dual-tree complex wavelet transform is that it can be used to implement 2D wavelet transforms that are more selective with respect to orientation than is the separable 2D DWT. Most of the natural images have short span high frequencies and low frequencies extending for larger span. Hence, the design of our feature vector is such that it provides higher spatial localization and lower frequency resolution at higher frequencies and the reverse for lower frequencies. The energy and mean of the frequency content of the image at various sub bands and different spatial resolution (higher for higher frequency bands) is stored as feature vector. Thus, the given feature vector encodes high frequency information as well.

A method to simulate inhomogeneously irradiated objects with a superposition of 1D models

In close binary systems the atmosphere of one or both components can be significantly influenced by irradiation from the companion. Often the irradiated atmosphere is simulated with a single-temperature approximation for the entire half-sphere. We present a scheme to take the varying irradiation angle into account by combining several separate 1D models. This is independent of the actual code which provides the separate stellar spectra. We calculate the projected area of zones with given irradiation angle and use this geometrical factor to scale separate 1D models. As an example we calculate two different irradiation scenarios with the PHOENIX code. The scheme to calculate the projected area is applicable independent of the physical mechanism that forms these zones. In the case of irradiation by a primary with T=125000 K, the secondary forms ions at different ionisation states for different irradiation angles. No single irradiation angle exists which provides an accurate description of the spectrum. We show a similar simulation for weaker irradiation, where the profile of the H$\alpha$ line depends on the irradiation angle.