Various Forms Of Noise Research Articles

Alaryngeal Speech Enhancement for Noisy Environments Using a Pareto Denoising Gated LSTM

Loss of the larynx significantly alters natural voice production, requiring alternative communication modalities and rehabilitation methods to restore speech intelligibility and improve the quality of life of affected individuals. This paper explores advances in alaryngeal speech enhancement to improve signal quality and reduce background noise, focusing on individuals who have undergone laryngectomy. In this study, speech samples were obtained from 23 Lithuanian males who had undergone laryngectomy with secondary implantation of the tracheoesophageal prosthesis (TEP). Pareto optimized gated LSTM was trained on tracheoesophageal speech data to recognize complex temporal connections and contextual information in speech signals. The system was able to distinguish between actual speech and various forms of noise and artifacts, resulting in a 25% drop in the mean signal-to-noise ratio (SNR) compared to other approaches. According to acoustic analysis, the system significantly decreased the number of unvoiced frames (PVF) from 40% to 10% while maintaining stable proportions of voiced frames (PVS) and average voicing evidence (AVE), indicating the accuracy of the approach in selectively attenuating noise and undesired speech artifacts while preserving important speech information.

Adversarial attacks and adversarial training for burn image segmentation based on deep learning.

Deep learning has been widely applied in the fields of image classification and segmentation, while adversarial attacks can impact the model's results in image segmentation and classification. Especially in medical images, due to constraints from factors like shooting angles, environmental lighting, and diverse photography devices, medical images typically contain various forms of noise. In order to address the impact of these physically meaningful disturbances on existing deep learning models in the application of burn image segmentation, we simulate attack methods inspired by natural phenomena and propose an adversarial training approach specifically designed for burn image segmentation. The method is tested on our burn dataset. Through the defensive training using our approach, the segmentation accuracy of adversarial samples, initially at 54%, is elevated to 82.19%, exhibiting a 1.97% improvement compared to conventional adversarial training methods, while substantially reducing the training time. Ablation experiments validate the effectiveness of individual losses, and we assess and compare training results with different adversarial samples using various metrics.

SwinIBSR: Towards real-world infrared image super-resolution

Traditional infrared image super-resolution (SR) methods often fall short in real-world scenarios, particularly when dealing with images degraded by various forms of noise and artifacts. Addressing this gap, this paper presents SwinIBSR, a blind SR approach optimized for real-world infrared imaging conditions. SwinIBSR introduces a practical degradation model, specifically designed to simulate the complex degradation processes encountered in real-world infrared imaging. Additionally, a mixed training strategy using both infrared and visible light datasets greatly enhances the network’s generalization capabilities, leading to the production of high-quality SR images. Building upon the Transformer-based architecture of SwinIR, SwinIBSR is adept at learning intricate mappings from low-resolution to high-resolution images. Our comprehensive experiments validate the effectiveness of SwinIBSR, showcasing its outstanding performance in processing real-world infrared images and its notable superiority over existing methods regarding visual quality. This work contributes to the field of infrared image SR, offering insights in enhancing SR techniques for practical applications.

PointCVaR: Risk-Optimized Outlier Removal for Robust 3D Point Cloud Classification

With the growth of 3D sensing technology, the deep learning system for 3D point clouds has become increasingly important, especially in applications such as autonomous vehicles where safety is a primary concern. However, there are growing concerns about the reliability of these systems when they encounter noisy point clouds, either occurring naturally or introduced with malicious intent. This paper highlights the challenges of point cloud classification posed by various forms of noise, from simple background noise to malicious adversarial/backdoor attacks that can intentionally skew model predictions. While there's an urgent need for optimized point cloud denoising, current point outlier removal approaches, an essential step for denoising, rely heavily on handcrafted strategies and are not adapted for higher-level tasks, such as classification. To address this issue, we introduce an innovative point outlier cleansing method that harnesses the power of downstream classification models. Using gradient-based attribution analysis, we define a novel concept: point risk. Drawing inspiration from tail risk minimization in finance, we recast the outlier removal process as an optimization problem, named PointCVaR. Extensive experiments show that our proposed technique not only robustly filters diverse point cloud outliers but also consistently and significantly enhances existing robust methods for point cloud classification. A notable feature of our approach is its effectiveness in defending against the latest threat of backdoor attacks in point clouds.

All-optical image denoising using a diffractive visual processor

Image denoising, one of the essential inverse problems, targets to remove noise/artifacts from input images. In general, digital image denoising algorithms, executed on computers, present latency due to several iterations implemented in, e.g., graphics processing units (GPUs). While deep learning-enabled methods can operate non-iteratively, they also introduce latency and impose a significant computational burden, leading to increased power consumption. Here, we introduce an analog diffractive image denoiser to all-optically and non-iteratively clean various forms of noise and artifacts from input images – implemented at the speed of light propagation within a thin diffractive visual processor that axially spans <250 × λ, where λ is the wavelength of light. This all-optical image denoiser comprises passive transmissive layers optimized using deep learning to physically scatter the optical modes that represent various noise features, causing them to miss the output image Field-of-View (FoV) while retaining the object features of interest. Our results show that these diffractive denoisers can efficiently remove salt and pepper noise and image rendering-related spatial artifacts from input phase or intensity images while achieving an output power efficiency of ~30–40%. We experimentally demonstrated the effectiveness of this analog denoiser architecture using a 3D-printed diffractive visual processor operating at the terahertz spectrum. Owing to their speed, power-efficiency, and minimal computational overhead, all-optical diffractive denoisers can be transformative for various image display and projection systems, including, e.g., holographic displays.

Dynamic visual noise has limited influence on the habitat selection and behavioural activity of crustaceans and cephalopods

AbstractEnvironments contain various forms of noise that can interfere with the ability of animal sensory systems to perceive information. One ubiquitous type of visual noise in shallow aquatic habitats is caustic flicker (or caustics), consisting of dynamically moving light patterns caused by the refraction of light when passing through the water's rippling surface. While some teleost fish avoid environments with caustic noise (where their prey can be more difficult to detect), it remains untested whether caustics affect the habitat selection of invertebrates. In the present study, we ask whether three invertebrate species, the shore crab Carcinus maenas, the brown shrimp Crangon crangon, and the common cuttlefish Sepia officinalis, prefer or avoid associating with environments with caustic noise, and whether caustics affect their behavioural activity and habitat exploration. To do this, we exposed the three species in binary choice experiments to different simulated caustic noise levels varying in their temporal (speed) and spatial (definition) components. Neither of the three tested invertebrate species spent more or less time in environments with higher caustic noise levels. While we also found no evidence that caustics affected the behavioural activity and exploration of Ca. maenas and S. officinalis, the brown shrimp Cr. crangon reduced its activity with increasing spatial caustic noise. However, all obtained effect sizes in this study were small, suggesting that caustic noise only minimally affects invertebrate behaviour. Overall, our results show that, unlike in teleost fish, caustics have limited influence on the habitat selection, exploration, and activity of crustaceans and cephalopods.

EEG Signal Filtering Method and Comparison Between Modern to Classical Filter

In neuroscience and therapeutic applications, electroencephalography (EEG) has become a crucial technique for gaining understanding of brain activity and cognitive processes. In order to improve signal quality, lower noise, and extract useful information from EEG data, preprocessing is essential. In the context of EEG signal processing, this study compares the effectiveness of three frequently used filtering methods: the median filter, average filter, and Gaussian filter.This study's main goal is to assess and compare how well these filters work to reduce the various forms of noise that may be found in EEG data, such as physiological artifacts, environmental interference, and electrode noise. The study makes use of a large dataset of EEG recordings that spans a range of experimental settings and mental states. The filtering methods are methodically applied to the unprocessed EEG data, and then the effects on signal fidelity and quality are thoroughly examined.The findings of this investigation shed important light on the advantages and disadvantages of each filtering strategy used in EEG data processing. In the end, the conclusions from this comparison study help offer criteria for choosing a suitable filtering approach based on the unique properties of EEG data and the research goals.The findings of this investigation offer important new understandings of the advantages and disadvantages of each filtering strategy used in EEG data processing. Finally, depending on the unique properties of the EEG data and the study goals, the findings from this comparison analysis help offer criteria for choosing an acceptable filtering approach.

Opening the envelope: Efficient envelope-based PPG denoising algorithm

Photoplethysmography (PPG) signals obtained from the skin’s surface offer valuable insights into blood volume fluctuations. With the rising interest in continuous non-invasive physiological monitoring, PPG has garnered significant attention. However, PPG signals are often affected by various forms of noise, impeding reliable feature extraction. Robust data pre-processing approaches are vital for both retrospective and real-time analysis. Existing denoising methods, including recent machine learning techniques, often suffer from implementation challenges, computational inefficiency, and limited interpretability. Addressing this challenge, we propose a novel PPG denoising algorithm. The algorithm was evaluated using a dataset representing approximately 81,015.99 min or 1360.27 h of PPG data collected from 31 patients. The evaluation involved the calculation and analysis of five key metrics: Signal-to-Noise Ratio (SNR), Variance, Total Variation (TV), Shannon entropy, and Instances-per-second (IPS). Our results demonstrate a notable increase in SNR after denoising, indicating effective noise reduction while preserving signal content. Variance and TV values showed a reduction post-denoising, suggesting smoother and less variable signals, validating the noise suppression efficacy. Additionally, Shannon entropy exhibited a decrease after denoising, indicating successful noise reduction and enhanced signal regularity. The nonparametric Wilcoxon signed-rank test (a = 0.05) was employed to assess the statistical significance of the observed differences of these metrics before and after denoising. Furthermore, the computational speed analysis revealed the EPDA’s potential for efficient processing of large datasets and real-time applications. This comprehensive evaluation approach allows for a thorough understanding of the EPDA’s effectiveness in denoising PPG data, fostering advancements in non-invasive physiological monitoring and promoting the broader adoption of PPG-based healthcare technologies.

Context-aware dynamic filtering network for confocal laser endomicroscopy image denoising

Objective. As an emerging diagnosis technology for gastrointestinal diseases, confocal laser endomicroscopy (CLE) is limited by the physical structure of the fiber bundle, leading to the inevitable production of various forms of noise during the imaging process. However, existing denoising methods based on hand-crafted features inefficiently deal with realistic noise in CLE images. To alleviate this challenge, we proposed context-aware kernel estimation and multi-scale dynamic fusion modules to remove realistic noise in CLE images, including multiplicative and additive white noise. Approach. Specifically, a realistic noise statistics model with random noise specific to CLE data is constructed and further used to develop a self-supervised denoised model without the participation of clean images. Secondly, context-aware kernel estimation, which improves the representation of features by similar learnable region weights, addresses the problem of the non-uniform distribution of noises in CLE images and proposes a lightweight denoised model (CLENet). Thirdly, we have developed a multi-scale dynamic fusion module that decouples and recalibrates features, providing a precise and contextually enriched representation of features. Finally, we integrated two developed modules into a U-shaped backbone to build an efficient denoising network named U-CLENet. Main Results. Both proposed methods achieve comparable or better performance with low computational complexity on two gastrointestinal disease CLE image datasets using the same training benchmark. Significance. The proposed approaches improve the visual quality of unclear CLE images for various stages of tumor development, helping to reduce the rate of misdiagnosis in clinical decision-making and achieve computer graphics-assisted diagnosis.

A Non-Invasive Circuit Breaker Arc Duration Measurement Method with Improved Robustness Based on Vibration–Sound Fusion and Convolutional Neural Network

Previous studies have shown that the contact wear estimation of circuit breakers can be based on the accumulative arc duration. However, one problem that remains unresolved is how to reliably measure the arc duration. Existing methods encounter difficulties in implementation and suffer from limited accuracy owing to the impact of the substation environment. To overcome these issues, this article presents a novel, non-invasive method for measuring arc duration that combines vibration–sound fusion and convolutional neural network. The proposed method demonstrates excellent performance, achieving errors below 0.1 ms under expected noise conditions and less than 1 ms in the presence of various forms of noise, transient interference, and even sensor failure. Its advantages include its ability to accurately measure arc duration and its robustness against noise and interference with unknown patterns and varying intensity as well as sensor failure. These features make it highly suitable for practical deployment in substation environments.

Large-scale chemical process causal discovery from big data with transformer-based deep learning

Fault diagnosis is critical for ensuring safe and stable chemical production. Correct identification of causal relationships among variables in large-scale chemical processes is a prerequisite for analyzing the root causes and propagation paths of faults. However, chemical process big data often exhibit nonlinearity and nonstationarity, and contain various forms of noise, rendering conventional causal discovery methods vulnerable. In this paper, a novel causal discovery method based on the causality-gated time series Transformer (CGTST) is proposed to address this challenge. By performing time series prediction using the Transformer-based model on the target variable, CGTST measures the causal strength by assessing the contribution of each variable to the prediction through the causality gate structure. Furthermore, a causal validation method based on permutation feature importance is proposed to eliminate spurious causal relationships and ensure robust results. To enhance the performance of causal discovery on nonlinear and nonstationary chemical process data, ensemble empirical mode decomposition is employed to reduce noise. The CGTST-based method is validated on three case studies: a continuous stirred-tank reactor, the Tennessee Eastman process, and a real-world continuous catalytic reforming process. Our findings demonstrate that the proposed method outperforms conventional causal discovery methods and holds promising prospects for industrial applications.

Revealing geometrically necessary dislocation density from electron backscatter patterns via multi-modal deep learning

The characterization of geometrically necessary dislocation (GND) is central to understanding the plastic deformation in materials. Currently, fast and accurate determination of GND density via Electron Backscatter Diffraction (EBSD) remains a challenge. Here, a multi-modal deep learning approach is proposed to predict GND density in terms of electron backscatter patterns (EBSPs) and dislocation configurations. The proposed multi-modal architecture consists of two separated convolutional neural network (CNN) processing streams. One CNN stream aims at extracting pattern shifts from EBSPs, and the other CNN stream focuses on learning suitable representations of dislocation configurations. We also introduce a specific data augmentation strategy termed neighboring pairs generating strategy for the GND prediction task. Taking the GND density from dictionary indexing-based analysis as the target property, high accuracy is achieved on several aluminum samples. Also, our networks are robust to various forms of noise, and the prediction speed is as fast as modern EBSD scanning rates, enabling real-time GND density analysis possible.

Denoising Deep Learning Network Based on Singular Spectrum Analysis—DAS Seismic Data Denoising With Multichannel SVDDCNN

Distributed acoustic sensing (DAS) is a new tool with low cost, sensitive signal capture, and complete coverage for vertical seismic profile (VSP) acquisition. Although DAS has obvious advantages over geophones, some weaknesses may limit its application. The main challenge is that DAS is polluted by various types of noise, including optical abnormal noise, random background noise, fading noise, and so on. To suppress these novel noises, we developed a new denoising neural network based on singular spectrum analysis—multichannel singular value decomposition denoising convolutional neural network (SVDDCNN). The network can simultaneously extract data features from singular spectrum instead of the time domain, which can represent geophysical features more accurately and help separate signals from noises. Second, a multichannel input layer is designed, and the input is decomposed into three subspaces by singular spectrum analysis, which provides records of different signal-to-noise ratios (SNRs) for training and improves generalization ability of the network. Third, to enhance the quality of the data set, we added the noise subspace records removed by SVD into the training set to provide various forms of noise with different singular spectra. Both synthetic and field examples show that our network has achieved impressive denoising of DAS VSP and demonstrated competitive performance compared with other methods. Furthermore, the structure similarity (SSIM) map is introduced to evaluate the signal leakage by calculating the similarity between the denoised record and the removed noise record. The lowest SSIM index of the proposed network indicated superior signal preservation ability.

About the features of the least squares method with linearization when determining nonlinear parameters of functional dependencies

The features of approximation of empirical data by functional dependence with nonlinear parameters using the two-stage least squares method are considered in this paper. A method of simplified parameter estimation by constructing a new expression that depends on the parameters in a linear way is described. To obtain the final solution, the least squares estimation of the main dependence linearized in terms of parameters is performed. The influence of various forms of noise imposed on the theoretical dependence on the approximations is modeled.

Research on Deep Convolutional Neural Network Time-Frequency Domain Seismic Signal Denoising Combined With Residual Dense Blocks

Deep Convolutional Neural Networks (DCNN) have the ability to learn complex features and are thus widely used in the field of seismic signal denoising with low signal-to-noise ratio (SNR). However, the current convolutional deep network used for seismic signal noise reduction does not make full use of the feature information extracted from all convolution layers in the network, and thus cannot fit the seismic signal with high SNR. To deal with this problem, this paper proposes the DnRDB model, a convolutional deep network time-frequency domain seismic signal denoising model combined with residual dense blocks (RDB). The model is mainly composed of several RDB in series. The input of each convolution layer in each RDB module is formed by the output of all the previous convolution layers. Meanwhile, even if the number of layers is increased, the fusion of the seismic signal features learned by the RDB modules can still achieve full extraction of seismic signals. Furthermore, deepening the model structure by concatenating multiple RDB modules enables further useful feature information to be extracted, which improves the SNR of seismic signals. The DnRDB model was trained and tested using the Stanford Global Seismic Dataset. The experimental results show that the DnRDB model can effectively recover seismic signals and remove various forms of noise. Even in the case of high noise, the denoised signal still has a high SNR. When the DnRDB model is compared with other denoising approaches such as wavelet threshold, empirical mode decomposition, and different deep learning methods, the results indicate that it performs best overall in denoising the same segment of the noisy seismic signal; the denoised signal also has less waveform distortion. Use of the DnRDB model in subsequent seismic signal processing work indicates that it can help the phase recognition algorithm improve the accuracy of seismic recognition through noise reduction.

Robust modelling of binary decisions in Laplacian Eigenmaps-based Echo State Networks

This paper aims to present a framework for supervised binary classification of n-Boolean functions through Echo State Networks endowed with Laplacian Eigenmaps for dimensionality reduction. The proposed method is applied both to improve the classification performance when the learnt weights are quantised in view of a digital implementation and as a computational demonstration of the neural reuse theory when parallel outputs are allowed. Our analysis focuses on the effect of various forms of noise (i.e., normal noise, uniform noise and quantisation noise) when all the possible Boolean functions of n input bits are learnt. External disturbances are applied both over the learnt weights and the input features so that we can analyse how resilient the whole architecture is when various forms of parametric noise is injected into the system. Results presented here show that dimensionality reduction allowed by the Laplacian Eigenmaps-based approach improves robustness to these different sources of noise, leading to reduced memory storage requirements while maintaining high classification performance. Our results are compared to those derived from other more common classification techniques in terms of learning performance and computational complexity, also considering a realistic dataset describing a decision making task in a wall-following navigation session with mobile robots.

Unknown noise removal via sparse representation model

To remove more complex or unknown noise, we propose a new dictionary learning model by assuming noise as Mixture of Gaussian (MoG) distributions. Since MoG is able to approximate any continuous distributions universally, the proposed method can effectively recover the original image from the corrupted one with various forms of noise. Meanwhile, to solve weighted ℓ2−ℓ0 minimization problems, we further propose modified orthogonal matching pursuit method in sparse coding and extend alternating proximal algorithm to update dictionaries. Experimental results demonstrate that our proposed method is superior to several previous denoising methods in terms of quantitative measures and visual quality.

A hypothesis study on bionic active noise reduction of auditory organs

BackgroundNoise exposure can lead to hearing loss and multiple system dysfunctions. As various forms of noise exist in our living environments, and our auditory organs are very sensitive to acoustic stimuli, it is a challenge to protect our hearing system in certain noisy environments.Presentation of the hypothesisHerein, we propose that our hearing organ could serve as a noise eliminator for high intensity noise and enhance acoustic signal processing abilities by increasing the signal-noise ratio. For suprathreshold signals, the hearing system is capable of regulating the middle ear muscles and other structures to actively suppress the sound level to a safe range.Testing the hypothesisTo test our hypothesis, both mathematic model analyses and animal model studies are needed. Based on a digital 3D reconstructed model, every structure in the auditory system can be analyzed and tested for its contribution to the process of noise reduction. Products manufactured by this bionic method could be used and verified in animal models and volunteers.ImplicationsBy mimicking the noise-reduction effect of the sophisticated structures in the hearing system, we may be able to provide a model that establishes a new active-sound-suppression mode. This innovative method may overcome the limited capabilities of current noise protection options and become a promising possibility for noise prevention.

Influence of Noise-Generating Factors on Cross-Correlation Electron Backscatter Diffraction (EBSD) Measurement of Geometrically Necessary Dislocations (GNDs)

Studies of dislocation density evolution are fundamental to improved understanding in various areas of deformation mechanics. Recent advances in cross-correlation techniques, applied to electron backscatter diffraction (EBSD) data have particularly shed light on geometrically necessary dislocation (GND) behavior. However, the framework is relatively computationally expensive-patterns are typically saved from the EBSD scan and analyzed offline. A better understanding of the impact of EBSD pattern degradation, such as binning, compression, and various forms of noise, is vital to enable optimization of rapid and low-cost GND analysis. This paper tackles the problem by setting up a set of simulated patterns that mimic real patterns corresponding to a known GND field. The patterns are subsequently degraded in terms of resolution and noise, and the GND densities calculated from the degraded patterns using cross-correlation ESBD are compared with the known values. Some confirmation of validity of the computational degradation of patterns by considering real pattern degradation is also undertaken. The results demonstrate that the EBSD technique is not particularly sensitive to lower levels of binning and image compression, but the precision is sensitive to Poisson-type noise. Some insight is also gained concerning effects of mixed patterns at a grain boundary on measured GND content.

ASEP of MIMO System with MMSE-OSIC Detection over Weibull-Gamma Fading Channel Subject to AWGGN

Ordered successive interference cancellation (OSIC) is adopted with minimum mean square error (MMSE) detection to enhance the multiple-input multiple-output (MIMO) system performance. The optimum detection technique improves the error rate performance but increases system complexity. Therefore, MMSE-OSIC detection is used which reduces error rate compared to traditional MMSE with low complexity. The system performance is analyzed in composite fading environment that includes multipath and shadowing effects known as Weibull-Gamma (WG) fading. Along with the composite fading, a generalized noise that is additive white generalized Gaussian noise (AWGGN) is considered to show the impact of wireless scenario. This noise model includes various forms of noise as special cases such as impulsive, Gamma, Laplacian, Gaussian, and uniform. Consequently, generalizedQ-function is used to model noise. The average symbol error probability (ASEP) of MIMO system is computed for 16-quadrature amplitude modulation (16-QAM) using MMSE-OSIC detection in WG fading perturbed by AWGGN. Analytical expressions are given in terms of Fox-H function (FHF). These expressions demonstrate the best fit to simulation results.