- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
Reconstruction of compressively sampled light field by using tensor dictionaries
摘要： How to capture the high quality light field photography was one of important issue in computational photography. In fact, light field could be captured directly for all views or compressively reconstructed for each view just through one coded image. The latter kind of method was more feasible since only one exposure was needed for all views, among which dictionary-based light field reconstruction had been shown its effectiveness. In this paper, a more effective light field reconstruction method based on tensor dictionary was created. The proposed method is efficient because the trained tensor form dictionary can make better use of the rich structure of light field. Specifically, multiple small dictionaries were trained at the same time, and then were combined to a big dictionary using Kronecker product. Experimental results demonstrate the proposed method outperforms a state-of-the-art reconstruction method with the vector-form dictionary, in terms of higher reconstruction PSNR while reducing the scale of dictionary substantially.
关键词： Dictionary learning,Compressive sensing,Tensor,Light field
更新于2025-09-23 15:19:57
-
[IEEE 2019 24th Microoptics Conference (MOC) - Toyama, Japan (2019.11.17-2019.11.20)] 2019 24th Microoptics Conference (MOC) - Wavelength Characteristics of a Silicon Waveguide Mach-Zehnder Interferometer having a Ce:YIG cladding
摘要： The problem of determining and understanding the nature of buried objects by means of nondestructive and non-invasive techniques represents an interesting issue for a great variety of applications. In this framework, the theory of electromagnetic inverse scattering problems can help in such an issue by starting from the measures of the scattered field collected on a surface. What will be presented in this communication is a two-dimensional (2-D) technique based on the so-called Born approximation (BA) combined with a compressive sensing (CS) approach, in order to improve reconstruction capabilities for a proper class of targets. The use of a multiview-multistatic configuration will be employed together with a multifrequency approach to overcome the limited amount of data due to the single-frequency technique. Therefore, after a first numerical analysis of the performance of the considered algorithm, some numerical examples for 2-D aspect-limited configurations will be presented. The scenario is composed of a simplified scene, which consists of two half-spaces, and with the probes located close to the interface between the two media. As proposed in the following, it is easy to observe that the use of CS for this kind of problems may improve reconstruction capabilities, confirming the validity of the presented approach.
关键词： microwave,scattering,ground penetrating,Compressive sensing (CS),tomography,inverse,electromagnetic radar
更新于2025-09-23 15:19:57
-
[IEEE 2019 IEEE/ACS 16th International Conference on Computer Systems and Applications (AICCSA) - Abu Dhabi, United Arab Emirates (2019.11.3-2019.11.7)] 2019 IEEE/ACS 16th International Conference on Computer Systems and Applications (AICCSA) - Multi-Layer Laser Scanner Strategy for Obstacle Detection and Tracking
摘要： A traf?c matrix is generally used by several network management tasks in a data center network, such as traf?c engineering and anomaly detection. It gives a ?ow-level view of the network traf?c volume. Despite the explicit importance of the traf?c matrix, it is signi?cantly dif?cult to implement a large-scale measurement to build an absolute traf?c matrix. Generally, the traf?c matrix obtained by the operators is imperfect, i.e., some traf?c data may be lost. Hence, we focus on the problems of recovering these missing traf?c data in this paper. To recover these missing traf?c data, we propose the spatio-temporal Kronecker compressive sensing method, which draws on Kronecker compressive sensing. In our method, we account for the spatial and temporal properties of the traf?c matrix to construct a sparsifying basis that can sparsely represent the traf?c matrix. Simultaneously, we consider the low-rank property of the traf?c matrix and propose a novel recovery model. We ?nally assess the estimation error of the proposed method by recovering real traf?c.
关键词： matrix completion,Traf?c matrix recovery,Kronecker compressive sensing,network management,network measurement
更新于2025-09-23 15:19:57
-
[IEEE 2019 IEEE PES Asia-Pacific Power and Energy Engineering Conference (APPEEC) - Macao, Macao (2019.12.1-2019.12.4)] 2019 IEEE PES Asia-Pacific Power and Energy Engineering Conference (APPEEC) - A Single-Stage Flyback LED Driver Based On Energy Distribution Without Electrolytic Capacitor
摘要： The lp (0 < p < 1) regularization has attracted a great attention in the compressive sensing field, because it can obtain sparser solutions than the well-known l1 regularization. Recently, we developed an approximate general analytic thresholding representation for any lp regularization with 0 < p < 1. The derived thresholding representations are exact for the well-known soft-threshold filtering for l1 regularization and the hard-threshold filtering for l0 regularization. Because the lp regularization is a nonconvex problem, an iterative algorithm can only converge to local optima instead of the global optimum. In this paper, we propose an alternating iteration algorithm for computed tomography reconstruction in a thresholding form based on our general analytic thresholding representation for better convergent properties. The alternating iteration algorithm alternatively minimizes one l1 and one lp (0 < p < 1) regularized objective functions. While the lp regularization can help to find a sparser solution, the l1 regularization can help to monitor the solution not away from the global optimum. Both numerical simulations and phantom experiments are performed to evaluate the proposed alternating iteration algorithm. Compared with the lp (0 < p < 1) regularization using a single p, the proposed alternating iteration algorithm reduces more data measurements for accurate reconstruction and is more robust for projection noise.
关键词： image reconstruction,Compressive sensing,least square solution,computed tomography,alternating iteration,lp regularization
更新于2025-09-23 15:19:57
-
[IEEE 2019 International Applied Computational Electromagnetics Society Symposium - China (ACES) - Nanjing, China (2019.8.8-2019.8.11)] 2019 International Applied Computational Electromagnetics Society Symposium - China (ACES) - A Band-Stop Plasmonic Filter Based on Spoof Surface Plasmon Polaritons
摘要： Direction of arrival (DOA) estimation from the perspective of sparse signal representation has attracted tremendous attention in past years, where the underlying spatial sparsity reconstruction problem is linked to the compressive sensing (CS) framework. Although this is an area with ongoing intensive research and new methods and results are reported regularly, it is time to have a review about the basic approaches and methods for CS-based DOA estimation, in particular for the underdetermined case. We start from the basic time-domain CS-based formulation for narrowband arrays and then move to the case for recently developed methods for sparse arrays based on the co-array concept. After introducing two specifically designed structures (the two-level nested array and the co-prime array) for optimizing the virtual sensors corresponding to the difference co-array, this CS-based DOA estimation approach is extended to the wideband case by employing the group sparsity concept, where a much larger physical aperture can be achieved by allowing a larger unit inter-element spacing and therefore leading to further improved performance. Finally, a specifically designed uniform linear array structure with associated CS-based underdetermined DOA estimation is presented to exploit the difference co-array concept in the spatio-spectral domain, leading to a significant increase in degrees of freedom. Representative simulation results for typical narrowband and wideband scenarios are provided to demonstrate their performance.
关键词： Compressive sensing,difference co-array,direction of arrival estimation,sparse array structures,underdetermined
更新于2025-09-23 15:19:57
-
[IEEE 2019 Conference on Lasers and Electro-Optics Europe & European Quantum Electronics Conference (CLEO/Europe-EQEC) - Munich, Germany (2019.6.23-2019.6.27)] 2019 Conference on Lasers and Electro-Optics Europe & European Quantum Electronics Conference (CLEO/Europe-EQEC) - Systematic Design of Photonic Crystal Cavities with Ultra-Low Modal Volume Considering Different Fabrication Resolutions
摘要： Sparsity of the ratings available in the recommender system database makes the task of rating prediction a highly underdetermined problem. This poses a limit on the accuracy and the quality of prediction. In this paper, we utilize secondary information pertaining to user’s demography and item categories to enhance prediction accuracy. Within the matrix factorization framework, we introduce additional supervised label consistency terms that match the user and item factor matrices to the available secondary information (metadata). Matrix factorization model—conventionally employed in collaborative ?ltering techniques—yields dense user and dense item factor matrices—the assumption is that users have an af?nity toward all latent factors and items possess all latent factors. Our formulation, based on a recent work, aims to recover a dense user and a sparse item factor matrix—this is a more reasonable model. Human beings show a natural interest toward all the factors, but every item cannot possess all the factors; this leads to a sparse item factor matrix. A natural outcome of our proposal is a solution to the pure cold start problem. We utilize the label consistency map generated from the proposed model to make reasonable recommendations for new users and new items which have not (been) rated yet. We demonstrate the performance of our model for a movie recommendation system. We also design an ef?cient algorithm for our formulation.
关键词： cold start,blind compressive sensing,matrix factorization,latent factor model,Auxiliary information
更新于2025-09-23 15:19:57
-
Analytical Model for Photonic Compressive Sensing With Pulse Stretch and Compression
摘要： Compressive sensing (CS) with photonic technologies provides a promising way to acquire information with reduced measurement. Photonic CS with pulse stretch and compression has been proved to be capable of capturing wideband time-domain signals at extremely high equivalent sampling rate or images at high frame rate. In this approach, an input short pulse is first stretched by propagating through a dispersive medium and then the stretched pulse is modulated by a signal to be measured and a pseudorandom bit sequence (PRBS). The stretched pulse encoded with the signal and the PRBS is compressed in the time domain after passing a second dispersive medium with an opposite dispersion value. The time-domain compression of the stretched pulse was regarded as the integration function in the CS process (but it has never been proved), which is a key to realize time-domain imaging. In this paper, we fully investigate the theoretical framework of the photonic CS with optical pulse stretch and compression, and present an analytical model of the CS measurement matrix based on the analysis of the pulse stretch, modulation and compression, for the first time to our knowledge. Moreover, we prove the equivalence between the peak value of the compressed pulse and the integral value of the mixed signal, which is the basis of the analytical model. In addition, we further discuss the impact of the limited bandwidth of the employed photodetector on the measurement and the performance of signal reconstruction.
关键词： microwave photonics,Compressive sensing
更新于2025-09-19 17:15:36
-
[IEEE 2018 IEEE Biomedical Circuits and Systems Conference (BioCAS) - Cleveland, OH, USA (2018.10.17-2018.10.19)] 2018 IEEE Biomedical Circuits and Systems Conference (BioCAS) - Block-Sparse Compressive Sensing for High-Fidelity Recording of Photoplethysmogram
摘要： This paper presents a novel compressive sensing (CS) framework for photoplethysmogram (PPG) recording. Exploiting the concept of block sparsity in CS, the proposed framework trains a block-sparsifying dictionary for the PPG signal using the block K-SVD (BK-SVD) algorithm. Next, the block sparse Bayesian learning (BSBL) algorithm is employed to utilize the block-sparsity information and recover the PPG signal from its compressively sampled counterpart. Using different PPG datasets prerecorded from the fingertip of a healthy human volunteer under normal and post-exercise conditions, our results demonstrate that the proposed CS framework based on BK-SVD + BSBL can achieve signal-to-noise and distortion ratio (SNDR) values of >10dB for compression ratios as high as 10, outperforming the previous approaches for compressive sensing of PPG that do not utilize the block-sparsity information.
关键词： photoplethysmogram,wearable health monitoring,Block sparsity,dictionary training,compressive sensing
更新于2025-09-19 17:15:36
-
[IEEE 2019 10th Workshop on Hyperspectral Imaging and Signal Processing: Evolution in Remote Sensing (WHISPERS) - Amsterdam, Netherlands (2019.9.24-2019.9.26)] 2019 10th Workshop on Hyperspectral Imaging and Signal Processing: Evolution in Remote Sensing (WHISPERS) - A Framework For An Artificial Neural Network Enabled Single Pixel Hyperspectral Imager
摘要： Compressive Sensing enables improvement of acquisition of a variety of signals in various applications with little to no discernible loss in terms of recovered image quality. The current work proposes a signal processing framework for the acquisition and fast reconstruction of compressively sampled hyperspectral images using an artificial neural network architecture. This ANN-based approach is capable of performing a fast reconstruction by avoiding the requirement of solving a computationally intensive image-specific optimization problem. The proposed framework contributes to advance single-pixel hyperspectral imaging device methodologies, which enable a significant reduction in device mechanical complexity, imaging rate, and cost. Our experiments demonstrate that a hyperspectral image can be reconstructed using only 10% of the samples without compromising classification performance. Specifically, the results show that classification performance of the compressively sampled hyperspectral image recovered using artificial neural networks is equal or higher to that of those obtained using current scanning hyperspectral imaging platforms.
关键词： remote sensing,deep learning,hyperspectral imaging,compressive sensing
更新于2025-09-19 17:13:59
-
[IEEE 2019 Conference on Lasers and Electro-Optics Europe & European Quantum Electronics Conference (CLEO/Europe-EQEC) - Munich, Germany (2019.6.23-2019.6.27)] 2019 Conference on Lasers and Electro-Optics Europe & European Quantum Electronics Conference (CLEO/Europe-EQEC) - Ultrafast Light Source at 1.8 ??m Based on Thulium-Doped Fibers for Three-Photon Microscopy
摘要： Optical tomographic imaging requires an accurate forward model as well as regularization to mitigate missing-data artifacts and to suppress noise. Nonlinear forward models can provide more accurate interpretation of the measured data than their linear counterparts, but they generally result in computationally prohibitive reconstruction algorithms. Although sparsity-driven regularizers significantly improve the quality of reconstructed image, they further increase the computational burden of imaging. In this paper, we present a novel iterative imaging method for optical tomography that combines a nonlinear forward model based on the beam propagation method (BPM) with an edge-preserving three-dimensional (3-D) total variation (TV) regularizer. The central element of our approach is a time-reversal scheme, which allows for an efficient computation of the derivative of the transmitted wave-field with respect to the distribution of the refractive index. This time-reversal scheme together with our stochastic proximal-gradient algorithm makes it possible to optimize under a nonlinear forward model in a computationally tractable way, thus enabling a high-quality imaging of the refractive index throughout the object. We demonstrate the effectiveness of our method through several experiments on simulated and experimentally measured data.
关键词： Optical phase tomography,beam propagation method,compressive sensing,total variation regularization,stochastic proximal-gradient,sparse reconstruction
更新于2025-09-19 17:13:59