Components In Frequency Domain Research Articles

ReFID: Reciprocal Frequency-aware Generalizable Person Re-identification via Decomposition and Filtering

Domain generalization of person re-identification aims to conduct testing across domains that have not been previously encountered, without utilizing target domain data during the training stage. As the number of source domains increases, the relationships between training samples become more complex. This can lead to domain-invariant features that include certain instance-level spurious correlations, which can impact the model’s ability to generalize further. To overcome this limitation, the Reciprocal Frequency-aware Generalizable Person Re-identification method is proposed in this article, which aims to utilize spectral feature correlation learning to transmit frequency component information and generate more discriminative hybrid features. A module called Bilateral Frequency Component-guided Attention is developed to help the network understand high-level semantic and texture information from various frequency features. Furthermore, to reduce the impact of noise from the frequency domain, this article proposes an innovative module called Fourier Noise Masquerade Filtering. This module enhances the portability of frequency domain components while simultaneously suppressing elements that do not contribute to generalization. Extensive experimental results on various datasets demonstrate that our method is effective and superior to the state-of-the-art methods.

Influence and the Control of Shield Tunneling on Ancient Tower Vibration

In recent years, to effectively mitigate the issue of urban traffic congestion, tunnels have become widely used new type of road. However, when the soil layer is excavated and disturbed, the resulting vibration affects the safety of the buildings above it, especially for ancient buildings that are very sensitive to vibrations. Taking the White Tower in Hangzhou as an example, in this paper, the vibration response propagation of large-diameter shield tunnel excavation to adjacent ancient towers is studied through field measurements. The vibration response of the unexcavated south line tunnel is predicted by using 3D numerical analysis software, and the optimal construction parameters are obtained. The study found that the frequency domain component of the White Tower bedrock vibration caused by shield tunneling is mainly 5–20 Hz, and the vibration response attenuates significantly beyond 40 m. Further, reducing the propulsive force can reduce the vibration response of the White Tower; therefore, controlling the propulsion and reducing the tunneling velocity can reduce the vibration response.

MM-Net: A MixFormer-Based Multi-Scale Network for Anatomical and Functional Image Fusion.

Anatomical and functional image fusion is an important technique in a variety of medical and biological applications. Recently, deep learning (DL)-based methods have become a mainstream direction in the field of multi-modal image fusion. However, existing DL-based fusion approaches have difficulty in effectively capturing local features and global contextual information simultaneously. In addition, the scale diversity of features, which is a crucial issue in image fusion, often lacks adequate attention in most existing works. In this paper, to address the above problems, we propose a MixFormer-based multi-scale network, termed as MM-Net, for anatomical and functional image fusion. In our method, an improved MixFormer-based backbone is introduced to sufficiently extract both local features and global contextual information at multiple scales from the source images. The features from different source images are fused at multiple scales based on a multi-source spatial attention-based cross-modality feature fusion (CMFF) module. The scale diversity of the fused features is further enriched by a series of multi-scale feature interaction (MSFI) modules and feature aggregation upsample (FAU) modules. Moreover, a loss function consisting of both spatial domain and frequency domain components is devised to train the proposed fusion model. Experimental results demonstrate that our method outperforms several state-of-the-art fusion methods on both qualitative and quantitative comparisons, and the proposed fusion model exhibits good generalization capability. The source code of our fusion method will be available at https://github.com/yuliu316316.

Monocular endoscopy images depth estimation with multi-scale residual fusion

BackgroundMonocular depth estimation plays a fundamental role in clinical endoscopy surgery. However, the coherent illumination, smooth surfaces, and texture-less nature of endoscopy images present significant challenges to traditional depth estimation methods. Existing approaches struggle to accurately perceive depth in such settings. MethodTo overcome these challenges, this paper proposes a novel multi-scale residual fusion method for estimating the depth of monocular endoscopy images. Specifically, we address the issue of coherent illumination by leveraging image frequency domain component space transformation, thereby enhancing the stability of the scene's light source. Moreover, we employ an image radiation intensity attenuation model to estimate the initial depth map. Finally, to refine the accuracy of depth estimation, we utilize a multi-scale residual fusion optimization technique. ResultsTo evaluate the performance of our proposed method, extensive experiments were conducted on public datasets. The structural similarity measures for continuous frames in three distinct clinical data scenes reached impressive values of 0.94, 0.82, and 0.84, respectively. These results demonstrate the effectiveness of our approach in capturing the intricate details of endoscopy images. Furthermore, the depth estimation accuracy achieved remarkable levels of 89.3 % and 91.2 % for the two models’ data, respectively, underscoring the robustness of our method. ConclusionsOverall, the promising results obtained on public datasets highlight the significant potential of our method for clinical applications, facilitating reliable depth estimation and enhancing the quality of endoscopy surgical procedures.

Hybrid Wideband Beamforming for Sum Spectral Efficiency Maximization in Millimeter-Wave Relay-Assisted Multiuser MIMO Cognitive Radio Networks

Relay-assisted hybrid beamforming plays an inevitable role in enhancing network coverage, transmission range, and spectral efficiency while simultaneously reducing hardware cost, power consumption, and hardware implementation complexity. This study investigates a cognitive radio network (CRN)-based hybrid wideband transceiver for millimeter-wave (mm-wave) decode-and-forward (DF) relay-assisted multiuser (MU) multiple-input multiple-output (MIMO) systems. It is worth mentioning that the underlying problem has not been addressed so far, which is a real motivation behind the proposed algorithm. The joint optimization of hybrid processing components and the constant amplitude constraints imposed by the analog beamforming solution make this problem non-convex and NP-hard. Furthermore, the analog beamformer common to all sub-carriers is another challenging aspect of the underlying problem. To derive the frequency-flat analog processing component in the radio frequency (RF) domain and frequency-dependent baseband processing matrices in the baseband domain, the original complicated problem is reformulated as two single-hop sum-rate maximization sub-problems. Taking advantage of this decomposition, the sum spectral efficiency is maximized through RF precoding and combining. On the other hand, the impact of interference among transmitted data streams and inter-user interference (IUI) is minimized via baseband processing matrices. Finally, computer simulations are conducted by changing system parameters, considering both perfect and imperfect channel state information (CSI). Simulation results demonstrate that the proposed algorithm achieves performance close to full-complexity precoding and outperforms other well-known hybrid beamforming techniques. Specifically, more than 95% efficiency is achieved with perfect CSI, and more than 90% efficiency is attained under the assumption of 30% error in the estimated channels.

A Rapid Forward for Grounded Long Wire Source Transient Electromagnetic Method

Aiming at enhancing the forward and inversion efficiency of grounded long wire source TEM (Transient ElectroMagnetic method), a new rapid layered medium forward method is proposed in the paper. This forward method is based on two points, one is to approximate the vertical magnetic component in frequency domain of a grounded long wire source by the response of several circular loops, and the other is to introduce the technique of "flow through" into the digital filtering calculation of sinusoidal transformation. The result of model tests shows that the calculation efficiency of proposed method is improved greatly (dozens of times) compared with that of the conventional method.

Noise adaptive filtering model integrating spatio-temporal feature for soft sensor

Due to uncertain production conditions and external disturbances, there are a lot of noise in the industrial process dataset. Traditional denoising models can only solve a single noise problem. However, the noise in industrial process data has diversity and inhomogeneity. Therefore, the noise adaptive filtering (NAF) model integrating spatio-temporal feature (ST) (ST-NAF) is proposed for the soft sensor, which achieves a balance between noise and data features to reduce the influence of noise on soft sensor results. The gated recurrent unit extracts temporal features of the industrial process data. Then, the dynamically gated graph attention extracts the interaction relationship between monitoring points to obtain spatio-temporal features. Moreover, the NAF converts time-domain features into frequency-domain features, and adaptively selects frequency-domain components with important information through the noise filtering matrix. The selected frequency-domain components are converted to the temporal domain to obtain the denoised spatio-temporal features. Finally, the self-attention learns the dynamic relationships between denoised spatio-temporal features and output variables to predict the key indicator. Compared with other denoising soft sensor models in a public dataset and an actual industrial process dataset, the ST-NAF achieves state-of-the-art results. And the prediction accuracy of the proposed model improved by 4.5% and 10.0% in terms of R2, respectively. Furthermore, the experimental results show that the ST-NAF can adaptively select useful information from industrial noise data to achieve accurate prediction of key indicators.

Identification of hydration stages: An innovative study of ultrasonic coda waves using integrated sensing element (ISE)

In this study, a piezoelectric-based integrated sensing element (ISE) is prepared and preliminary employed for monitoring the hydration process based on the dominant mechanism of acoustic impedance variation of cement and coda wave idea. The results of the signal for a period of 24 h show that three different stages (0–2 h/2–12 h/12–24 h) of the cement during hydration can be distinguished by the reflected coda wave combined with its frequency-domain component, while the first arrival waveform shows almost the same pattern due to the wave trajectory only in packing material. Moreover, the wavelet analysis of the coda wave can distinguish these three stages, in line with the results from the time-domain waveform, energy, and frequency-domain spectrum.

Energy loss and pressure fluctuation characteristics of coastal two-way channel pumping stations under the ultra-low head condition

The ultra-low head condition is a unique condition of the coastal two-way channel pumping stations, but the quantitative research on its hydraulic characteristics still needs to be improved. The entropy production method with computational fluid dynamics and the high-frequency pressure fluctuation test were employed to analyze the pumping device's internal energy loss and pressure fluctuation. The results show that the primary source of energy loss of the two-way channel pumping device under the ultra-low head condition is the turbulent dissipation caused by the velocity fluctuation of the impeller and guide vane, in which there are large irregular high entropy production areas on the pressure surface of the impeller and in the impeller channel. The high-frequency pressure fluctuation in the pumping device is mainly concentrated near the impeller, and the amplitude under the ultra-low head is 3.5 times the amplitude under the design head. Due to the dynamic and static interaction between the impeller and the guide vane, the frequency domain components of the pressure fluctuation at the outlet channel bell mouth are complex. Under the ultra-low head condition, the pressure fluctuation degree is intensified, the energy loss of each part is greater, and the low-frequency resonance should be prevented.

Study on sinusoidal estimation deviation of electrostatic actuated MEMS mirror torsion angle

Electrostatic MEMS micromirrors usually work in resonant state to obtain large amplitude of torsion angle. The real-time prediction of MEMS micromirror torsion angle is calculated according to the measured resonant amplitude and phase under the assumption that the relationship between the torsion angle and time is sinusoidal. However, there are few reports on the deviation of this torsion angle predication based on sinusoidal assumption. In this paper, the real resonant torsion trajectory of C1100 MEMS micromirror under different driving frequencies and voltages is measured by using microscopic laser Doppler method, and the deviation between the real trajectory and the trajectory fitted by sinusoidal curve is compared. The results show that the real trajectory of the MEMS micromirror driven by square wave is not completely consistent with the sinusoidal estimation, and the deviation increases with the increase of the torsional angle amplitude. By obtaining the frequency domain components of the torsion angle signal using FFT method, the main reason of this prediction deviation is due to composition of harmonic signals on base frequency signal. The research results reveal that the sinusoidal assumption method is only suitable for situations when the optical angle accuracy is less than 0.1°.

Prediction of wave force on netting under strong nonlinear wave action

Under the strong nonlinear wave environment, accurate simulation of wave force for aquaculture netting is an effective guarantee for cage design and safety. In this paper, the horizontal wave forces of a nylon square-mesh netting panel were obtained through a series of strong nonlinear regular wave tests, and their nonlinearity was analyzed by amplitude spectrum. Moreover, the Morison equation based on fifth-order Stokes wave theory was used to reasonably predict the wave force on the netting. The results showed that both wave and wave force have strong nonlinearity, especially the latter. The frequency domain characteristics of the test wave and wave force are similar, while the higher frequency components of the test force are more apparent. The predicted wave forces are in good agreement with the test values in time and frequency domain, and zero or higher frequency components of predicted force are more prominent with the increase of wave steepness. When the range of the Keulegan-Carpenter number is 35-120, the average drag and inertia coefficient of the predicted force are 2.4 and 2.1, respectively. The results can provide a more accurate assessment of the nonlinear wave force on aquaculture facilities.

A Block-Scale FFT Filter Based on Spatial Autocorrelation Features of Speckle Noise in SAR Image

In order to reduce the impact of noise on the accuracy of inversion products based on SAR images, many filtering algorithms have been developed for noise reduction of SAR images. This paper proposes a filtering method based on the spatial autocorrelation feature of the block fast Fourier transform (BFFT). The method statistically analyses the autocorrelation length of speckle noise on Sentinel-1B images for different features and then constructs a relationship between autocorrelation length and noise period. After that, the size of the optimal FFT filtering window radius was determined based on the relationship between the noise period and the components in the image frequency domain. Finally, we filtered the SAR image within the parcels. We compared BFFT with six commonly used filtering methods. The results show that: (1) The noise periods of the soybean, corn, paddy, and water objects on the SAR image have little difference, with noise periods of 3.36, 3.17, 3.13, and 3.14 pixels on the VV polarization and 3.49, 3.17, 2.94, and 2.42 pixels on the VH polarization; (2) after the BFFT filtering in the land parcel area, the mean value of the backscattering coefficient (BC) kept constant, whilst at the same time, the standard deviation (STD) was reduced to half of that before the filtering and (3) the BFFT and NLM filtering methods have a better effect on noise reduction inside the block. The BFFT filtering method retains the variation trend between different regions within the block and preserves the block boundary’s clarity. This study provides a new idea for refined image processing.

Variational time–frequency adaptive decomposition of machine multi-impact vibration signals

The vibration signal of a mechanical equipment shell displays multi-impact coupling characteristics, when reciprocating mechanical vibration signals of large marine diesel engines and petrochemical reciprocating compressors, which exhibits the characteristics of nonstationarity, strong coupling, and strong interference in the time–frequency domain. Signal decomposition and multi-source impact identification have great significance for early fault diagnosis. However, existing decomposition methods for extracting fault frequency bands cannot satisfy demands for extracting impacts from the multi-source impact signals of reciprocating machinery. In this study, variational time–frequency adaptive decomposition (VTFAD) for multi-impact vibration signals is established to decompose and extract single-impact components. A variational model based on the minimization of a spectral center moment is proposed based on the two-dimensional morphological characteristics of energy concentration and discrete distribution in the time–frequency spectrum of the impact signal. An iterative solution is obtained using the alternating direction multiplier method; the time–frequency center and effective time–frequency domain components of the impact are calculated adaptively. A parameter optimization selection scheme is designed for selecting the number of time–frequency domain decompositions and quadratic penalty factor in the optimization process. The test of simulated mechanical multi-impact signals demonstrated that the proposed method can adaptively and accurately identify the time–frequency centers and time–frequency domain characteristics of the impacts in multi-impact vibration signals. VTFAD was applied to a diesel engine connecting the rod-bearing wear-fault vibration signal. The results show that VTFAD can effectively identify the new impacts caused by the fault and accurately extract the key characteristics of the impacts.

Synchronous analyses between electroencephalogram and surface electromyogram based on motor imagery and motor execution.

The functional coupling of the cerebral cortex and muscle contraction indicates that electroencephalogram (EEG) and surface electromyogram (sEMG) signals are coherent. The objective of this study is to clearly describe the coupling relationship between EEG and sEMG through a variety of analysis methods. We collected the EEG and sEMG data of left- or right-hand motor imagery and motor execution from six healthy subjects and six stroke patients. To enhance the coherence coefficient between EEG and sEMG signals, the algorithm of EEG modification based on the peak position of sEMG signals is proposed. Through analyzing a variety of signal synchronization analysis methods, the most suitable coherence analysis algorithm is selected. In addition, the wavelet coherence analysis method based on time spectrum estimation was used to study the linear correlation characteristics of the frequency domain components of EEG and sEMG signals, which verified that wavelet coherence analysis can effectively describe the temporal variation characteristics of EEG-sEMG coherence. In the task of motor imagery, the significant EEG-sEMG coherence is mainly in the imagination process with the frequency distribution of the alpha and beta frequency bands; in the task of motor execution, the significant EEG-sEMG coherence mainly concentrates before and during the task with the frequency distribution of the alpha, beta, and gamma frequency bands. The results of this study may provide a theoretical basis for the cooperative working mode of neurorehabilitation training and introduce a new method for evaluating the functional state of neural rehabilitation movement.

Resting state electroretinography: An innovative approach to intrinsic retinal function monitoring.

The electroretinogram (ERG) represents the biopotential evoked by the retina in response to a light stimulus. The flash evoked ERG (fERG) is the ERG modality most frequently used clinically to diagnose and monitor retinal disorders. We hereby present a new method to record spontaneous retinal activity, without the use of a flash stimulus, that we named the resting-state ERG (rsERG). The recordings were done in normal subjects under light- and dark-adaptation and with different background light conditions (i.e., variations of wavelength and intensity). Additionally, rsERG recordings were obtained in five patients with retinopathies. The signals were subsequently analyzed in the frequency domain, extracting both periodic (i.e., frequency peaks) and aperiodic (i.e., background trend) components of the signal. The later was further assessed through a multifractal analysis using Wavelet Leaders. Results show that, irrespective of the recording conditions used, the rsERG always includes the same 90 Hz component; a frequency component also present in the fERG response, suggesting a retinally-intrinsic origin. However, in addition, the fERGs also includes a low-frequency component which is absent in the rsERGs, a finding supporting a retinally-induced origin. Comparing rsERGs with fERGs in selected patients with various retinal disorders indicates that the two retinal signals are not always similarly affected (either as a result of underlying retinal pathology or otherwise), suggesting an added value in the assessment of retinal function. Thus, the rsERG could have a similar role in clinical visual electrophysiology as that of the resting-state EEG in neurology namely, to quantify changes in spontaneous activity that result from a given disease processes.

Frequency-domain segmentation algorithm for three-dimensional light-field display based on pixel deviation threshold

The three-dimensional (3D) light-field display technology with full-parallax is an inevitable future development trend, however, still suffered from pixel deviation effect. Although depth-priority arrangement and fitting algorithm have been investigated to address it, both have caused a decrease in the resolution or sharpness, respectively. In this paper, a frequency-domain segmentation algorithm based on the pixel deviation threshold is demonstrated to reconstruct high sharpness and the resolution of 3D light-field images without pixel deviation. This algorithm segments the 3D images in references to the pixel deviation threshold and combines the backward ray-tracing algorithm and the fitting algorithm to process the various frequency-domain components. Meanwhile, pixel deviation threshold with distinct depths is proposed and analyzed, and the simulation experiments verify the validity of the pixel deviation thresholds. Simulation and experimental results demonstrate that our proposed algorithm addresses the pixel deviation issue and exhibits improved performance in other aspects compared with the current algorithm.

Arid and semiarid rangeland responses to non-stationary temporal dynamics of environmental drivers

Temporal ecology is a growing research field aimed at understanding how temporal dynamics structure ecological systems in a changing environment. Ecosystems are exposed to accelerating environmental changes —mainly driven by climate change— that challenge their future survival. The main feature of this planetary process is that statistical properties such as mean, variance and autocorrelation are not constant over time. While this non-stationary climate change characteristic is well known, our premise is that far too much attention is given to ecosystems stationary responses, where annual cycles and seasonal changes of vegetation productivity are the main indicators for addressing environmental changes and planning rangeland management. Hence, shifts in other relevant intra-annual and inter-annual cycles are overlooked. We emphasize that a rigorous and systematic tool aimed at addressing these kinds of temporal changes in vegetation dynamics is lacking. We proposed an operative framework aimed at tackling the ability of rangelands to develop non-stationary responses in short periods, in the face of non-stationary changes in the dynamics of external factors driving its temporal functioning. In particular, we measured non-stationary dynamics of rangeland productivity by means of shifts in the hierarchy of frequency domain components over time of different rangelands’ photosynthetic activity from North-West Patagonia, Argentina, as measured by the Normalized Difference Vegetation Index (NDVI). We emphasize the importance of a relevant temporal dimension of ecosystems with new insights into both the understanding of environmental changes impact on terrestrial ecosystems and the implementation of rangelands adaptive management.

Data-Wise Spatial Regional Consistency Re-Enhancement for Hyperspectral Image Classification

Effectively using rich spatial and spectral information is the core issue of hyperspectral image (HSI) classification. The recently proposed Diverse Region-based Convolutional Neural Network (DRCNN) achieves good results by weighted averaging the features extracted from several predefined regions, thus exploring the use of spatial consistency to some extent. However, such feature-wise spatial regional consistency enhancement does not effectively address the issue of wrong classifications at the edge of regions, especially when the edge is winding and rough. To improve the feature-wise approach, Data-wise spAtial regioNal Consistency re-Enhancement (“DANCE”) is proposed. Firstly, the HSIs are decomposed once using the Spectral Graph Wavelet (SGW) to enhance the intra-class correlation. Then, the image components in different frequency domains obtained from the weight map are filtered using a Gaussian filter to “debur” the non-smooth region edge. Next, the reconstructed image is obtained based on all filtered frequency domain components using inverse SGW transform. Finally, a DRCNN is used for further feature extraction and classification. Experimental results show that the proposed method achieves the goal of pixel level re-enhancement with image spatial consistency, and can effectively improve not only the performance of the DRCNN, but also that of other feature-wise approaches.

A neuronal theta band signature of error monitoring during integration of facial expression cues.

Error monitoring is the metacognitive process by which we are able to detect and signal our errors once a response has been made. Monitoring when the outcome of our actions deviates from the intended goal is crucial for behavior, learning, and the development of higher-order social skills. Here, we explored the neuronal substrates of error monitoring during the integration of facial expression cues using electroencephalography (EEG). Our goal was to investigate the signatures of error monitoring before and after a response execution dependent on the integration of facial cues. We followed the hypothesis of midfrontal theta as a robust neuronal marker of error monitoring since it has been consistently described as a mechanism to signal the need for cognitive control. Also, we hypothesized that EEG frequency-domain components might bring advantage to study error monitoring in complex scenarios as it carries information from locked and non-phase-locked signals. A challenging go/no-go saccadic paradigm was applied to elicit errors: integration of facial emotional signals and gaze direction was required to solve it. EEG data were acquired from twenty healthy participants and analyzed at the level of theta band activity during response preparation and execution. Although theta modulation has been consistently demonstrated during error monitoring, it is still unclear how early it starts to occur. We found theta power differences at midfrontal channels between correct and error trials. Theta was higher immediately after erroneous responses. Moreover, before response initiation we observed the opposite: lower theta preceding errors. These results suggest theta band activity not only as an index of error monitoring, which is needed to enhance cognitive control, but also as a requisite for success. This study adds to previous evidence for the role of theta band in error monitoring processes by revealing error-related patterns even before response execution in complex tasks, and using a paradigm requiring the integration of facial expression cues.

Parsimonious truncated Newton method for time-domain full-waveform inversion based on the Fourier-domain full-scattered-field approximation

To exploit Hessian information in full-waveform inversion (FWI), the matrix-free truncated Newton method can be used. In such a method, Hessian-vector product computation is one of the major concerns due to the huge memory requirements and demanding computational cost. Using the adjoint-state method, the Hessian-vector product can be estimated by zero-lag crosscorrelation of the first-/second-order incident wavefields and the second-/first-order adjoint wavefields. Different from the implementation in frequency-domain FWI, Hessian-vector product construction in the time domain becomes much more challenging because it is not affordable to store all of the time-dependent wavefields. The widely used wavefield recomputation strategy leads to computationally intensive tasks. We have developed an efficient alternative approach to computing the Hessian-vector product for time-domain FWI. In our method, discrete Fourier transform is applied to extract frequency-domain components of involved wavefields, which are used to compute wavefield crosscorrelation in the frequency domain. This makes it possible to avoid reconstructing the first- and second-order incident wavefields. In addition, a full-scattered-field approximation is proposed to efficiently simplify the second-order incident and adjoint wavefield computation, which enables us to refrain from repeatedly solving the first-order incident and adjoint equations for the second-order incident and adjoint wavefields (re)computation. With our method, the computational time can be reduced by 70% and 80% in viscous media for Gauss-Newton and full-Newton Hessian-vector product construction, respectively. The effectiveness of our method is also verified in the frame of a 2D multiparameter inversion, in which our method almost reaches the same iterative convergence of the conventional time-domain implementation.