Accurate Reconstruction Method Research Articles

Multitask Deep Learning Reconstruction and Localization of Lesions in Limited Angle Diffuse Optical Tomography.

Diffuse optical tomography (DOT) leverages near-infrared light propagation through tissue to assess its optical properties and identify abnormalities. DOT image reconstruction is an ill-posed problem due to the highly scattered photons in the medium and the smaller number of measurements compared to the number of unknowns. Limited-angle DOT reduces probe complexity at the cost of increased reconstruction complexity. Reconstructions are thus commonly marred by artifacts and, as a result, it is difficult to obtain an accurate reconstruction of target objects, e.g., malignant lesions. Reconstruction does not always ensure good localization of small lesions. Furthermore, conventional optimization-based reconstruction methods are computationally expensive, rendering them too slow for real-time imaging applications. Our goal is to develop a fast and accurate image reconstruction method using deep learning, where multitask learning ensures accurate lesion localization in addition to improved reconstruction. We apply spatial-wise attention and a distance transform based loss function in a novel multitask learning formulation to improve localization and reconstruction compared to single-task optimized methods. Given the scarcity of real-world sensor-image pairs required for training supervised deep learning models, we leverage physics-based simulation to generate synthetic datasets and use a transfer learning module to align the sensor domain distribution between in silico and real-world data, while taking advantage of cross-domain learning. Applying our method, we find that we can reconstruct and localize lesions faithfully while allowing real-time reconstruction. We also demonstrate that the present algorithm can reconstruct multiple cancer lesions. The results demonstrate that multitask learning provides sharper and more accurate reconstruction.

SemAGR: semantic method for accurate geolocations reconstruction within extensive urban sites

Accurate geolocation of users’ posts in social networks plays a vital role in a wide area of research devoted to analysing the urban environment based on social media data. It allows us to effectively correlate information about a real urban object with its description in social networks. Information analysis about extensive urban sites is challenging in such studies because publications correlate extremely unevenly with marked geotags relating to such objects. In addition, users of social networks often intentionally indicate incorrect geolocations to increase their publications’ popularity. This article proposes a solution to the problem of accurate geolocation reconstruction for extensive urban sites. It is proposed to use a combined method that enriches the list of initial geolocations of the social network with the help of external sources. The transformer model is then applied to recognise named entities that help to detect mentions of events or locations in user posts. After detection, posts are redefined to the new locations from the extended list according to the semantic similarity calculated between publications and locations.

Accurate floorplan reconstruction using geometric priors

We present an accurate and automatic bottom-up floorplan reconstruction method by leveraging geometric priors extracted from raw point clouds of indoor scenes. Compared to two state-of-the-art methods which adopt point density as priors only, our designed geometric priors integrate point density with indoor area recognition and normal information. These geometric priors are used to calculate the confidence score for each unit region as part of the external boundaries. A cost function is developed according to the confidence scores and the normals along a certain edge, as well as the edge length. By minimizing the cost function, we can automatically estimate candidate edges, and connect them edge by edge to generate rough external boundaries. After diminishing the redundant edges, we can finally solve for a fine closed-loop external boundary path via an efficient heuristic method. We validate our method on intensive real indoor scenes. Experimental results show that our method outperforms two state-of-the-art methods in terms of reconstructing accurate external floorplan boundaries.

IPPG 2 cPPG: Reconstructing contact from imaging photoplethysmographic signals using U-Net architectures

Imaging photoplethysmography (iPPG) is an optical technique dedicated to the assessment of several vital functions using a simple camera. Significant efforts have been made to reliably estimate heart and respiratory rates. Currently, research is focusing on the remote estimation of oxygen saturation and blood pressure (BP). The limited number of publicly available data tends to restrict the advancements related to BP estimation. To overcome this limit, we propose to split the problem in a two-stage processing chain: (i) converting iPPG to contact PPG (cPPG) signals using available video dataset and (ii) estimate BP from converted cPPG signals by exploiting large existing databases (e.g. MIMIC). This article presents the first developments where a method for converting iPPG signals measured using a camera into cPPG signals measured by contact sensors is proposed. Real and imaginary parts of the continuous wavelet transform (CWT) of cPPG and iPPG signals are passed to various deep pre-trained U-shaped architectures. Conventional metrics and specific waveform estimators have been implemented to validate the relevance of the predictions. The results exhibit good agreements towards a large portion of metrics, showing that the neural architectures properly estimated cPPG from iPPG signals through their CWT representations. The performance indicates that BP estimation from iPPG signals converted to cPPG signals can now be envisaged. Consequently, future work will focus on the integration of models dedicated to BP estimation trained on MIMIC. This is the first demonstration of a method for accurate reconstruction of cPPG from iPPG signals satisfying pulse waveform criteria.

Field reconstruction and off-design performance prediction of turbomachinery in energy systems based on deep learning techniques

Obtaining the off-design characteristics of core components such as turbines and compressors is the basis of off-design analysis for energy systems. However, the characteristics are difficult to accurately acquire in the initial design stage of turbomachinery. Based on deep learning techniques, an accurate and rapid field reconstruction and off-design aerodynamic performance prediction method is proposed. First, a Generative Adversarial Network with the added Bezier layer is employed to establish a database of turbine blade profiles. Then, a Dual Convolutional Neural Network (Dual-CNN) is established to reconstruct the pressure and temperature fields as well as predict the off-design performances of different profiles and working conditions. Based on the above two kinds of neural networks, a turbine in a solar-based supercritical carbon dioxide Brayton cycle is taken as an example. The field reconstruction and off-design performance prediction are conducted on the basis of the established rotor blade profile database. The accuracy of field reconstruction is guaranteed. The off-design performance prediction of the established Dual-CNN indicates that the example blade profile is suitable for operation with larger mass flow rate. Compared with the traditional method, Dual-CNN can reduce the off-design analysis time of one blade geometry from 38.4 h to 7.68s.

An application of the edge reversal method for accurate reconstruction of the three-dimensional profile of a single-point diamond tool obtained by an atomic force microscope

An application of the edge reversal method for accurate reconstruction of the three-dimensional profile of a single-point diamond tool obtained by an atomic force microscope

A modified simulated annealing algorithm for hybrid statistical reconstruction of heterogeneous microstructures

Mathematical descriptors of complex microstructures easily obtained from limited microstructural information offer next-generation, inexpensive solutions for digital material reconstruction. In this work, we propose a versatile method for accurate digital reconstruction and design of material’s microstructure. The method uses a combination of the full set of two-point correlation function (TPCF) and two-point correlation cluster function (TPCCF) applicable to two-dimensional (2D) and three-dimensional (3D) reconstructions. We employ a modified Simulated annealing (SA) algorithm to accelerate the optimization procedure using a novel probability-based neighborhood generator consisting of three different neighborhood functions. We demonstrate the applications of the method in reconstruction of solid oxide fuel cells (SOFC) and similar complex-microstructure heterogeneous materials. In each case, we demonstrate the accuracy of the method using several distinctive statistical measures, namely TPCF, TPCCF, porosity, and specific surface in 2D and 3D spaces.

GRMT: Generative Reconstruction of Mutation Tree From Scratch Using Single-Cell Sequencing Data.

Single-cell sequencing (SCS) now promises the landscape of genetic diversity at single cell level, and is particularly useful to reconstruct the evolutionary history of tumor. There are multiple types of noise that make the SCS data notoriously error-prone, and significantly complicate tumor tree reconstruction. Existing methods for tumor phylogeny estimation suffer from either high computational intensity or low-resolution indication of clonal architecture, giving a necessity of developing new methods for efficient and accurate reconstruction of tumor trees. We introduce GRMT (Generative Reconstruction of Mutation Tree from scratch), a method for inferring tumor mutation tree from SCS data. GRMT exploits the k-Dollo parsimony model to allow each mutation to be gained once and lost at most k times. Under this constraint on mutation evolution, GRMT searches for mutation tree structures from a perspective of tree generation from scratch, and implements it to an iterative process that gradually increases the tree size by introducing a new mutation per time until a complete tree structure that contains all mutations is obtained. This enables GRMT to efficiently recover the chronological order of mutations and scale well to large datasets. Extensive evaluations on simulated and real datasets suggest GRMT outperforms the state-of-the-arts in multiple performance metrics. The GRMT software is freely available at https://github.com/qasimyu/grmt.

Three-dimensional cellwise conservative unsplit geometric VOF schemes

This work presents two unsplit geometric VOF schemes that extend the two-dimensional cellwise conservative unsplit (CCU) scheme (Comminal et al. (2015) [49]) to three dimensions. The novelty of the 3D-CCU schemes lies in the representation of the streaksurfaces of donating regions by polyhedral surfaces whose vertices are calculated with the 4th order Runge-Kutta scheme. Moreover, the advected liquid volumes are computed using a truncation algorithm (López et al. (2019) [62]) suited for arbitrary non-convex and self-intersecting polyhedra, which removes the need for tetrahedral decomposition. The 3D-CCU advection schemes were coupled to three interface reconstruction methods (Youngs' method, the Mixed Youngs-Centered scheme, and the Least-Square Fit algorithm). The resulting VOF methods were tested in classical benchmark advection tests, including translation, rigid-body rotation, shear and deformation flows. The proposed 3D-CCU schemes conserve the liquid volume and maintain the physical boundedness of liquid volume fractions to the machine precision. The 3D-CCU schemes perform favorably compared to other unsplit geometric VOF schemes when coupled to Youngs' interface reconstruction method. Moreover, the 3D-CCU schemes coupled to the Least-Square Fit algorithm are more accurate than most other VOF schemes that use a second-order accurate interface reconstruction, except those where a 3D extension of the Mosso-Swartz interface reconstruction is employed. The comparison of the different VOF schemes highlights the importance of coupling accurate interface reconstruction methods with accurate unsplit advection schemes.

Accurate reconstruction for the measurement of tilt surfaces with digital holography

Digital holography has been widely applied in the measurement of complex micro–nano structures. The reconstruction of the object wavefront is usually under the assumption that the measured sample is parallel to the recording plane, which means the diffraction distance is fixed. While the tilt of the measured sample is inevitable, especially for the scanning measurement of a large component where the movement will introduce inclination. Then ringing artifacts will occur, which seriously affect the quality of the reconstructed wavefronts. In this paper, an accurate diffractive reconstruction method is proposed. A pose transformation model is established for calculating the accurate diffraction distances of the tilt wavefronts. The gradient of the reconstructed complex amplitude is utilized in the merit function for the pose estimation. By minimizing the merit function, the pose parameters can be refined with the particle swarm optimization. Experimental results demonstrate that the proposed method can effectively improve the reconstruction accuracy of the tilt wavefronts. It is significant for the scanning measurement of large components.

Three-Dimensional Reconstruction of Dilute Bubbly Flow Field With Light-Field Images Based on Deep Learning Method

The three-dimensional reconstruction of bubble flow field is of great significance to study the motion of gas-liquid two phase flow. Combined with the abundant information of bubbles in the light field images, a fast and accurate 3D reconstruction method for dilute bubbly flow based on the deep learning algorithm DIF-LeNet (Double Information Fusion with LeNet5 net) is proposed in this work. The calibration method and bubble segmentation algorithm of light field image for data processing is detailed. DIF-LeNet realizes the fusion of different dimension data and regression prediction. By DIF-LeNet, bubble depth could be predicted by fusing the refocused bubble image with the focal distance of the refocused image. Then, the three-dimensional bubble field could be reconstructed with the bubble depth, bubble center and bubble diameter. Comparing with the reconstruction method based on statistics and LeNet5, the reconstruction accuracy and speed of the proposed method based on DIF-LeNet are improved. Especially for the bubble without focused image in the refocused image sequence, the effect of 3D reconstruction based on DIF-LeNet is much prominent.

A New Embedded Analysis with Pinwise Discontinuity Factors for Pin Power Reconstruction

Based on embedded analysis, an accurate pin power reconstruction (PPR) method is proposed for conventional nodal analysis. Unlike the common form function (FF) method, the new PPR method, named the embedded pin power reconstruction (EPPR) method, directly solves a two-group fixed-source problem that is defined with pinwise homogenized group constants (HGCs) and coarse-mesh incoming partial currents on the boundary. In the EPPR scheme, the pinwise HGCs including the pinwise discontinuity factor are predetermined from single-assembly lattice calculations, and the boundary partial currents are obtained from two-step nodal analyses. Two EPPR approaches are proposed: One is a 3×3 extended color-set configuration, and the other is a smaller one considering the half-thickness of the surrounding fuel assemblies. The performance of the EPPR methods is evaluated with various benchmark problems including partially mixed oxide–loaded pressurized water reactor cores, and the results are compared with the conventional FF method. Comprehensive results of this work demonstrate that the new EPPR method can provide much better accuracy than the conventional FF-based PPR method.

Accurate reconstruction of polarization parameters for channeled spectroscopic Stokes polarimeters.

In this work, we present an accurate polarization reconstruction method based on the coherence demodulation technique, which is different from the previous windowing method operating in the optical path difference domain. The proposed method uses a signal multiplier and a low-pass filter to reconstruct Stokes parameters without performing any Fourier transform. Because this method does not require a Fourier transform, the Stokes reconstruction could be finished in the spectral domain. For calibrating the waveplate phase error, coherence demodulation allows for establishing an analytical model to describe the influence of waveplate imperfections on the polarization measurement process. The phase error will result in a channel shift and Fourier broadening, both of which cause serious errors during Stokes reconstruction. With the model, a method based on a linear polarizer was proposed for calibrating the phase deviation of waveplate. After that, the accurate reconstruction of polarization parameters could be achieved. An experiment was performed to check the ability of the proposed method. The experimental result showed that it has the same excellent performance of reconstructing Stokes parameters using the traditional windowing method. Finally, a series of simulations was carried out to verify the robustness of this method, which showed that the reconstruction technique is robust to misalignment and additional noise.

Shape sensing of multilayered composite and sandwich beams based on Refined Zigzag Theory and inverse finite element method

An accurate and effective reconstruction method, which is commonly known as “shape sensing”, is presented for real-time reconstructing the deformation field of multilayered composite and sandwich laminated beam using the measured surface strains is presented. The proposed strategy uses the inverse finite element (iFEM) as a general framework and the Refine Zigzag Theory (RZT) as improved theory. Compared to the traditional iFEM, an additional kinematic variable is introduced to improve reconstruction accuracy of the axial displacement through the depth of the beam. Initially, the displacement fields, strain fields and equilibrium relations are presented. Then, the deformed displacement fields are described by the isogeometric analysis method, where the order of kinematic variables can be determined based on the equilibrium relations and the criteria for element interpolations. Next, transformation relationships are established between measured surface strains and section strains. And the number of strain sensor is further reduced through the boundary conditions. Finally, a least-square function is employed to establish the strain–displacement model. A cantilevered beam is used as a case to compare the reconstructed displacements with the theoretical displacement solutions. The numerical results demonstrate the superior accuracy capability and potential applicability of presented approach for performing accurate shape prediction of the RZT beam.

A More Accurate Reconstruction Method for Detecting Large-Depth Defects in Plates Using SH Guided Waves

The inspection of thickness thinning defects and corrosion defects is greatly significant for the health prediction of plate structures. The main aim of this research is to propose a novel and effective approach to achieve the accurate and rapid detection of arbitrary defects using shear horizontal (SH) guided waves, particularly for large-depth and complex defects. The proposed approach combines the quantitative detection of Fourier transform with a reference model-based strategy to improve the accuracy of large-depth defect detection. Since the shallow defect profile is theoretically constructed by inverse Fourier transform of the product of reflection coefficients and integral coefficients of reference models, the unknown large-depth defect can be initially assessed using the relevant information from a predefined reference model. By iteratively updating the integral coefficients of reference models, the accuracy of reconstruction of large-depth defects is much improved. To achieve the converged defect profile, a termination criterion, the root mean square error, is applied to guarantee the construction of defects with a high level of accuracy. Moreover, the hybrid finite element method is used to simulate the propagation of SH guided waves in plates for calculating the reflection coefficients of plates with defects. Finally, to demonstrate the capability of the developed reconstruction method for defect detection in terms of accuracy and efficiency, three types of large-depth defect profiles, i.e., a rectangular flaw, a double-rectangular flaw and a complex flaw, are examined. Results show that the discrepancy between the predicted defect profile and the real one is quite small, even in the largest-depth defect case where the defect depth is equal to 0.733 times the plate thickness, the minimal difference is observed. It is noted that the fast convergence of the proposed approach can be achieved by no more than ten updates for the worst case.

Human Body Shape Reconstruction from Binary Image Using Convolutional Neural Network

Human body modeling is an important part of virtual try-on. In order to quickly reconstruct the three-dimensional(3D) human body based on the minimum input, we propose a new method for accurate reconstruction of the human body by inputting binary images from either single view or multiple views. We first encode the shape of the human body via Principal Component Analysis (PCA) to extract the low dimensional shape descriptor. Secondly, we design a novel Body Reconstruction Convolutional Neural Network (BRCNN) with two branches, which could capture deep correlated features from different views and merge them. Given the obtained statistical shape space of the human body, we jointly train the BRCNN to learn a global mapping from the input to the shape descriptor which can be then decoded to points cloud for the reconstruction of various body shapes under neutral poses. The experimental results show that compared with the existing human reconstruction technology, the accuracy has been improved by 1.07%, and the prediction results of the two views are better than those from the single view. Further investigation also reveals that the prediction results of the weight-sharing network are better than the network without weight-sharing.

An accurate interface reconstruction method using piecewise circular arcs

SummaryA novel piecewise circular interface construction (PCIC) method for accurate reconstruction of interface in a two‐phase flow problem is proposed. This is under the framework of a fixed grid, volume of fluid approach applied on a two‐dimensional semistaggered structured grid. Fluid interface in each mixed cell is represented using a geometric template of piecewise circular arc. Data corresponding to arc center coordinates and radius are first predicted using curve fitting methods and corrected with the help of volume fraction constraints. Further corrections are carried out to achieve function (c0) continuity at cell boundaries. The proposed method does not require additional calculations for the determination of curvature (for calculation of surface tension force), since it is obtained as part of reconstruction process itself. For dynamic interface construction, simple analytical expressions are derived to construct edge matched flux polygons. Area of intersection of flux polygons with area covered by primary fluid is determined to effect geometric advection across a PCIC interface. Accuracy of this method is demonstrated by the reconstruction of standard static and dynamically evolving interface problems. Accuracy levels superior to most interface reconstruction methods using PLIC and schemes using higher order curves are established. Finally, the capability to handle a complex two‐phase flow problem simulation viz the four‐vortex flow field, where interface undergoes breakage and coalescence, is also demonstrated.

Efficient L1/2 Regularization-Based Reconstruction for Photoacoustic Imaging Using Adaptively Iterative Thresholding Algorithm

Sparse sampling photoacoustic imaging (PAI) is a significant research topic. Accurate and efficient reconstruction methods play important roles in the wide application of PAI in preclinical research. Compressed sensing has large potential in efficiently reconstructing high quality PAI images with sparse sampling signal. We presented our study on a L1/2 regularization operator based adaptive iterative thresholding method for sparse view PAI in this article, which can use few measurements to exactly reconstruct the PAI image. The effectiveness of the proposed algorithm was verified by simulation analysis using Matlab. And the proposed algorithm was compared with two state-of-the-art L1 norm and TV norm based algorithms for signal recovery and image reconstruction. The SNR of the reconstructed image calculated by the introduced method was better than the other two methods tested in this research. The algorithm presented in this paper not only provides higher quality image with sparse view data, but also is robust to noise and over estimation of sparsity value.

Fast reconstruction of atomic-scale STEM-EELS images from sparse sampling

This paper discusses the reconstruction of partially sampled spectrum-images to accelerate the acquisition in scanning transmission electron microscopy (STEM). The problem of image reconstruction has been widely considered in the literature for many imaging modalities, but only a few attempts handled 3D data such as spectral images acquired by STEM electron energy loss spectroscopy (EELS). Besides, among the methods proposed in the microscopy literature, some are fast but inaccurate while others provide accurate reconstruction but at the price of a high computation burden. Thus none of the proposed reconstruction methods fulfills our expectations in terms of accuracy and computation complexity. In this paper, we propose a fast and accurate reconstruction method suited for atomic-scale EELS. This method is compared to popular solutions such as beta process factor analysis (BPFA) which is used for the first time on STEM-EELS images. Experiments based on real as synthetic data will be conducted.

Moment-of-fluid analytic reconstruction on 3D rectangular hexahedrons

The moment-of-fluid method (MOF) is a second-order accurate interface reconstruction method which can be seen as an extension of the volume-of-fluid method with piecewise linear interface construction (VOF-PLIC). MOF involves a computationally intensive minimization problem that needs to be solved on every cell containing several materials. We propose a new fast and robust reconstruction algorithm to tackle this problem on rectangular hexahedral cells. Our approach uses explicit analytic formulas of the objective function that does not use any geometric computations such as half-space–polyhedron intersections. The numerical results show that the proposed method is more robust and more than 200 times faster than the original approach. Additionally, we propose a faster reconstruction algorithm on convex polyhedral cells. All the methods presented in this article have been implemented and verified on the open-source code Notus.