Search Results (14)

As one of the most crucial parts of face detection, the accuracy and the generalization of face anti-spoofing are particularly important. Therefore, it is necessary to propose a multi-branch network to improve the accuracy and generalization of the detection of unknown spoofing attacks. These branches consist of several frequency map encoders and one depth map encoder. These encoders are trained together. It leverages multiple frequency features and generates depth map features. High-frequency edge texture is beneficial for capturing moiré patterns, while low-frequency features are sensitive to color distortion. Depth maps are more discriminative than RGB images at the visual level and serve as useful auxiliary information. Supervised Multi-view Contrastive Learning enhances multi-view feature learning. Moreover, a two-stage feature fusion method effectively integrates multi-branch features. Experiments on four public datasets, namely CASIA-FASD, Replay–Attack, MSU-MFSD, and OULU-NPU, demonstrate model effectiveness. The average Half Total Error Rate (HTER) of our model is 4% (25% to 21%) lower than the Adversarial Domain Adaptation method in inter-set evaluations. Full article

Biosensors based on ion-sensitive field effect transistors (ISFETs) combined with aptamers offer a promising and convenient solution for point-of-care testing applications due to the ability for fast and label-free detection of a wide range of biomarkers. Mobile and easy-to-use readout devices for the ISFET aptasensors would contribute to further development of the field. In this paper, the development of a portable PC-controlled device for detecting aptamer-target interactions using ISFETs is described. The device assembly allows selective modification of individual ISFETs with different oligonucleotides. Ta₂O₅-gated ISFET structures were optimized to minimize trapped charge and capacitive attenuation. Integrated CMOS readout circuits with linear transfer function were used to minimize the distortion of the original ISFET signal. An external analog signal digitizer with constant voltage and superimposed high-frequency sine wave reference voltage capabilities was designed to increase sensitivity when reading ISFET signals. The device performance was demonstrated with the aptamer-driven detection of troponin I in both reference voltage setting modes. The sine wave reference voltage measurement method reduced the level of drift over time and enabled a lowering of the minimum detectable analyte concentration. In this mode (constant voltage 2.4 V and 10 kHz 0.1Vp-p), the device allowed the detection of troponin I with a limit of detection of 3.27 ng/mL. Discrimination of acute myocardial infarction was demonstrated with the developed device. The ISFET device provides a platform for the multiplexed detection of different biomarkers in point-of-care testing. Full article

Background: It is assumed that speech comprehension deficits in background noise are caused by age-related or acquired hearing loss. Methods: We examined young, middle-aged, and older individuals with and without hearing threshold loss using pure-tone (PT) audiometry, short-pulsed distortion-product otoacoustic emissions (pDPOAEs), auditory brainstem responses (ABRs), auditory steady-state responses (ASSRs), speech comprehension (OLSA), and syllable discrimination in quiet and noise. Results: A noticeable decline of hearing sensitivity in extended high-frequency regions and its influence on low-frequency-induced ABRs was striking. When testing for differences in OLSA thresholds normalized for PT thresholds (PTTs), marked differences in speech comprehension ability exist not only in noise, but also in quiet, and they exist throughout the whole age range investigated. Listeners with poor speech comprehension in quiet exhibited a relatively lower pDPOAE and, thus, cochlear amplifier performance independent of PTT, smaller and delayed ABRs, and lower performance in vowel-phoneme discrimination below phase-locking limits (/o/-/u/). When OLSA was tested in noise, listeners with poor speech comprehension independent of PTT had larger pDPOAEs and, thus, cochlear amplifier performance, larger ASSR amplitudes, and higher uncomfortable loudness levels, all linked with lower performance of vowel-phoneme discrimination above the phase-locking limit (/i/-/y/). Conslusions: This study indicates that listening in noise in humans has a sizable disadvantage in envelope coding when basilar-membrane compression is compromised. Clearly, and in contrast to previous assumptions, both good and poor speech comprehension can exist independently of differences in PTTs and age, a phenomenon that urgently requires improved techniques to diagnose sound processing at stimulus onset in the clinical routine. Full article

In this paper, a frequency discrimination method using a multiplicative-integral and linear transformation network is proposed. In this method, two preset differential frequency signals and frequency modulation signals are transformed by multiplication and integration, and then the instantaneous frequency parameters of the frequency modulation signal are accurately analyzed by the linear transformation network to restore the original modulation signal. Compared with the phase discriminator, the simulation results show that this method has a higher frequency discrimination bandwidth. In addition, this method has better anti-noise performance, and the frequency discrimination distortion caused by noise with a different Signal-to-Noise Ratio is reduced by 33.80% on average compared with the phase discriminator. What is more, the carrier center frequency error has little influence on the frequency discrimination quality of this method, which solves the problem that most common frequency discriminators are seriously affected by the carrier center frequency error. This method requires a low accuracy of carrier center frequency, which makes it extremely suitable for digital frequency discrimination technology and can meet the needs of various frequency discrimination occasions. Full article

In indoor environments, reverberation can distort the signalseceived by active noise cancelation devices, posing a challenge to sound classification. Therefore, we combined three speech spectral features based on different frequency scales into a densely connected network (DenseNet) to accomplish sound classification with reverberation effects. We adopted the DenseNet structure to make the model lightweight A dataset was created based on experimental and simulation methods, andhe classification goal was to distinguish between music signals, song signals, and speech signals. Using this framework, effectivexperiments were conducted. It was shown that the classification accuracy of the approach based on DenseNet and fused features reached 95.90%, betterhan the results based on other convolutional neural networks (CNNs). The size of the optimized DenseNet model is only 3.09 MB, which is only 7.76% of the size before optimization. We migrated the model to the Android platform. The modified model can discriminate sound clips faster on Android thanhe network before the modification. This shows that the approach based on DenseNet and fused features can dealith sound classification tasks in different indoor scenes, and the lightweight model can be deployed on embedded devices. Full article

Underwater optical imaging devices are often affected by the complex underwater environment and the characteristics of the water column, which leads to serious degradation and distortion of the images they capture. Deep learning-based underwater image enhancement (UIE) methods reduce the reliance on physical parameters in traditional methods and have powerful fitting capabilities, becoming a new baseline method for UIE tasks. However, the results of these methods often suffer from color distortion and lack of realism because they tend to have poor generalization and self-adaptation capabilities. Generating adversarial networks (GANs) provides a better fit and shows powerful capabilities on UIE tasks. Therefore, we designed a new network structure for the UIE task based on GANs. In this work, we changed the learning of the self-attention mechanism by introducing a trainable weight to balance the effect of the mechanism, improving the self-adaptive capability of the model. In addition, we designed a feature extractor based on residuals with multi-level residuals for better feature recovery. To further improve the performance of the generator, we proposed a dual path discriminator and a loss function with multiple weighted fusions to help model fitting in the frequency domain, improving image quality. We evaluated our method on the UIE task using challenging real underwater image datasets and a synthetic image dataset and compared it to state-of-the-art models. The method ensures increased enhancement quality, and the enhancement effect of the model for different styles of images is also relatively stable. Full article

The complex underwater environment and light scattering effect lead to severe degradation problems in underwater images, such as color distortion, noise interference, and loss of details. However, the degradation problems of underwater images bring a significant challenge to underwater applications. To address the color distortion, noise interference, and loss of detail problems in underwater images, we propose a triple-branch dense block-based generative adversarial network (TDGAN) for the quality enhancement of underwater images. A residual triple-branch dense block is designed in the generator, which improves performance and feature extraction efficiency and retains more image details. A dual-branch discriminator network is also developed, which helps to capture more high-frequency information and guides the generator to use more global content and detailed features. Experimental results show that TDGAN is more competitive than many advanced methods from the perspective of visual perception and quantitative metrics. Many application tests illustrate that TDGAN can significantly improve the accuracy of underwater target detection, and it is also applicable in image segmentation and saliency detection. Full article

Emotional intelligence (EI) is a critical social intelligence skill that refers to an individual’s ability to assess their own emotions and those of others. While EI has been shown to predict an individual’s productivity, personal success, and ability to maintain positive relationships, its assessment has primarily relied on subjective reports, which are vulnerable to response distortion and limit the validity of the assessment. To address this limitation, we propose a novel method for assessing EI based on physiological responses—specifically heart rate variability (HRV) and dynamics. We conducted four experiments to develop this method. First, we designed, analyzed, and selected photos to evaluate the ability to recognize emotions. Second, we produced and selected facial expression stimuli (i.e., avatars) that were standardized based on a two-dimensional model. Third, we obtained physiological response data (HRV and dynamics) from participants as they viewed the photos and avatars. Finally, we analyzed HRV measures to produce an evaluation criterion for assessing EI. Results showed that participants’ low and high EI could be discriminated based on the number of HRV indices that were statistically different between the two groups. Specifically, 14 HRV indices, including HF (high-frequency power), lnHF (the natural logarithm of HF), and RSA (respiratory sinus arrhythmia), were significant markers for discerning between low and high EI groups. Our method has implications for improving the validity of EI assessment by providing objective and quantifiable measures that are less vulnerable to response distortion. Full article

Reflective semiconductor optical amplifiers (RSOAs) are key elements for modern optical communications. Despite their widespread deployment, their performance when intended for ultrafast data amplification is limited by their inherently slow gain dynamics. In this paper, we propose to employ a birefringent fiber loop (BFL) to compensate for the RSOA pattern-dependent behavior and extend its operation well beyond that allowed by its nominal optical modulation bandwidth. We apply a reduced model to describe the RSOA response and quantify the RSOA output distortion by means of a non-return-to-zero data pulse overshoot. We validate the outcomes of this model in the time domain both for the RSOA alone and with the assistance of the BFL by an extensive comparison to available measurements. The excellent matching between simulation and experimental results allows us to further investigate the impact of critical operating parameters and derive specifications for them so that the performance of the scheme against the overshoot is made acceptable. The theoretical predictions confirm the ability of the BFL to enhance the RSOA direct amplification capability and hence establish it as a frequency discriminator for complementing RSOAs’ versatile and scalable operation. Full article

The increasing use of nonlinear loads in the power grid introduces some unwanted effects, such as harmonic and interharmonic contamination. Since the existence of spectral contamination causes waveform distortion that may be harmful to the loads that are connected to the grid, it is important to identify the frequency components that are related to specific loads in order to determine how relevant their contribution is to the waveform distortion levels. Due to the diversity of frequency components that are merged in an electrical signal, it is a challenging task to discriminate the relevant frequencies from those that are not. Therefore, it is necessary to develop techniques that allow performing this selection in an efficient way. This paper proposes the use of spectral kurtosis for the identification of stationary frequency components in electrical signals along the day in a sustainable building. Then, the behavior of the identified frequencies is analyzed to determine which of the loads connected to the grid are introducing them. Experimentation is performed in a sustainable building where, besides the loads associated with the normal operation of the building, there are several power electronics equipment that is used for the electric generation process from renewable sources. Results prove that using the proposed methodology it is possible to detect the behavior of specific loads, such as office equipment and air conditioning. Full article

The monitoring of human activity and vital signs plays a significant role in remote health-care. Radar provides a non-contact monitoring approach without privacy and illumination concerns. However, multiple people in a narrow indoor environment bring dense multipaths for activity monitoring, and the received vital sign signals are heavily distorted with body movements. This paper proposes a framework based on Frequency Modulated Continuous Wave (FMCW) and Impulse Radio Ultra-Wideband (IR-UWB) radars to address these challenges, designing intelligent spatial-temporal information fusion for activity and vital sign monitoring. First, a local binary pattern (LBP) and energy features are extracted from FMCW radar, combined with the wavelet packet transform (WPT) features on IR-UWB radar for activity monitoring. Then the additional information guided fusing network (A-FuseNet) is proposed with a modified generative and adversarial structure for vital sign monitoring. A Cascaded Convolutional Neural Network (CCNN) module and a Long Short Term Memory (LSTM) module are designed as the fusion sub-network for vital sign information extraction and multisensory data fusion, while a discrimination sub-network is constructed to optimize the fused heartbeat signal. In addition, the activity and movement characteristics are introduced as additional information to guide the fusion and optimization. A multi-radar dataset with an FMCW and two IR-UWB radars in a cotton tent, a small room and a wide lobby is constructed, and the accuracies of activity and vital sign monitoring achieve 99.9% and 92.3% respectively. Experimental results demonstrate the superiority and robustness of the proposed framework. Full article

Frequency response analysis (FRA) is a powerful and widely used tool for condition assessment in power transformers. However, interpretation schemes are still challenging. Studies show that FRA data can be influenced by parameters other than winding deformation, including temperature. In this study, a machine-learning approach with temperature as an input attribute was used to objectively identify faults in FRA traces. To the best knowledge of the authors, this has not been reported in the literature. A single-phase transformer model was specifically designed and fabricated for use as a test object for the study. The model is unique in that it allows the non-destructive interchange of healthy and distorted winding sections and, hence, reproducible and repeatable FRA measurements. FRA measurements taken at temperatures ranging from −40 °C to 40 °C were used first to describe the impact of temperature on FRA traces and then to test the ability of the machine learning algorithms to discriminate between fault conditions and temperature variation. The results show that when temperature is not considered in the training dataset, the algorithm may misclassify healthy measurements, taken at different temperatures, as mechanical or electrical faults. However, once the influence of temperature was considered in the training set, the performance of the classifier as studied was restored. The results indicate the feasibility of using the proposed approach to prevent misclassification based on temperature changes. Full article

Continuous monitoring of mechanical impacts is one of the goals of modern SHM systems using a sensor network installed on a structure. For the evaluation of the impact position, there are generally applied triangulation techniques based on the estimation of the differential time of arrival (DToA). The signals generated by impacts are multimodal, dispersive Lamb waves propagating in the plate-like structure. Symmetrical S₀ and antisymmetrical A₀ Lamb waves are both generated by impact events with different velocities and energies. The discrimination of these two modes is an advantage for impact positioning and characterization. The faster S₀ is less influenced by multiple path signal overlapping and is also less dispersive, but its amplitude is generally 40–80 dB lower than the amplitude of the A₀ mode. The latter has an amplitude related to the impact energy, while S₀ amplitude is related to the impact velocity and has higher frequency spectral content. For these reasons, the analog front-end (AFE) design is crucial to preserve the information of the impact event, and at the same time, the overall signal chain must be optimized. Large dynamic range ADCs with high resolution (at least 12-bit) are generally required for processing these signals to retrieve the DToA information found in the full signal spectrum, typically from 20 kHz to 500 kHz. A solution explored in this work is the design of a versatile analog front-end capable of matching the different types of piezoelectric sensors used for impact monitoring (piezoceramic, piezocomposite or piezopolymer) in a sensor node. The analog front-end interface has a programmable attenuator and three selectable configurations with different gain and bandwidth to optimize the signal-to-noise ratio and distortion of the selected Lamb wave mode. This interface is realized as a module compatible with the I/O of a 16 channels real-time electronic system for SHM previously developed by the authors. High-frequency components up to 270 kHz and lower-frequency components of the received signals are separated by different channels and generate high signal-to-noise ratio signals that can be easily treated by digital signal processing using a single central unit board with ADC and FPGA. Full article

The electric power systems together with the entire energy sector are rapidly evolving towards a low-carbon, secure, and competitive economy facing revolutionary transformations from technical structure to economic value chain. Pathways to achieve sustainability led to the development of new technologies, accommodation of larger shares of unpredictable and stochastic electricity transfer from sources to end-users without loss of reliability, new business models and services, data management, and so on. The new technologies and incentives for local energy communities along with large development of microgrids are main forces driving the evolution of the low voltage energy sector changing the context and paradigm of rigid contractual binding between utilities and end-user customers (now progressing to flexible prosumers with generation and storage capabilities). The flexibility and operation of a prosumer can be enhanced by a non-intrusive time-frequency analysis of distorted power quality waveforms for both generation and demand at the point of common connection. Therefore, it becomes of importance to discriminate among successive quasi-steady-state operation of a given local system using only the aggregated waveforms information available in the PCC. This paper focuses on the Hilbert–Huang method with modifications such as empirical mode decomposition improved with masking signals based on the Fast Fourier Transform, Hilbert spectral analysis, and a post-processing method for separating components and their amplitudes and frequencies within distorted power signals for a low-voltage prosumer operation. The method is used for a time-frequency-magnitude representation with promising localization capabilities enabling efficient operation for prosumers. Full article