Reduction Of Residual Noise Research Articles

System Residual Noise Reduction for Non-Contact Passive Intermodulation Measurements of Array Antennas Using Multiple Testers

System Residual Noise Reduction for Non-Contact Passive Intermodulation Measurements of Array Antennas Using Multiple Testers

A hybrid ICEEMDAN and OMEDA-based vibrodiagnosis method for the bearing of rolling stock

Purpose. To develop a hybrid method based on ICEEMDAN and OMEDA for high-fidelity detection of the diagnostic features of the technical condition of roller bearing. Methodology. Due to digital signal processing the search was made of the informative components on the broadband spectra, envelope spectra, squared envelope spectra. The usage of the methods of mathematical statistics for the selection of informative IMFs as a result of iterative calculate procedures. The introduction of blind deconvolution method with a proper objective function for the enhancement of impulse fault features. Findings. The use of traditional spectral methods provided an incomplete list of diagnostic features of bearing roller damage. The empirical mode decomposition (EMD) method and its minor versions provide the series of intrinsic mode functions (IMFs). On the broadband vibration spectra of the selected low-frequency IMFs, diagnostic features were not detected in a sufficient number. Further research focused on the development of a hybrid method that takes into account the demodulation procedure of the high-frequency components of those IMFs that were selected according to the complex evaluation index. The Fast Kurtogram technique determined an informative frequency band of 7.2–7.5 kHz for demodulation and construction of the squared envelope spectra, where harmonics of the roller spin frequency and some side bands with a width equal to the separator rotation frequency appeared. The use of OMEDA caused the reduction of residual noise on the squared envelope spectra and bilateral sidebands to appear around all harmonics of the roller spin frequency. Originality. For the first time, the whole list of the diagnostic features of roller faults of the axle-box bearing was identified using the developed hybrid method, which involves steps of noise reduction for enhancing a physical meaning of a proper frequency components, associated with faults. The usage of the optimal minimum entropy deconvolution adjusted method helped to eliminate noise and provide all diagnostic features of the technical condition on the squared envelope spectrum. Practical value. The results of the study help to control the quality of repair of the axle-box rolling bearings on the test rig at the workshop of the freight railway car depot.

Drone audition: Audio signal enhancement from drone embedded microphones using multichannel Wiener filtering and Gaussian-mixture based post-filtering

In this paper, we consider the problem of recovering desired sound source signals from on-board microphone recordings on a noisy drone. Enhancement of source signal degraded by drone noise is considered to be a difficult task due to the strong noise generated from its motors and propellers causing an extremely low signal-to-drone noise ratio (SD‾NR). We propose a solution (i) by combining the widely known multichannel Wiener filter (MWF) to remove drone noise from microphone recordings, and (ii) further reduction of residual noise using a Gaussian mixture model (GMM) based dual-stage parametric Wiener filter (WF). The method exploits known statistics of motor current-specific drone noise. This combination of techniques to the specific context of signal enhancement for drone audition is applicable to irregular microphone arrays embedded on a drone enabling realistic integration to most drones. We demonstrate the validity of the proposed framework with extensive real data through (i) experimental recordings from two different drone acoustics datasets and (ii) outdoor measurements from a hovering drone for a bioacoustic application. The results confirm improved performance in terms of SD‾NR, speech quality (PESQ), and intelligibility (STOI) at very low SD‾NR (up to −30 dB) and show a strong potential for signal enhancement applications using noisy drones.

Optimising weighted averaging for auditory brainstem response detection

The auditory brainstem response (ABR) is a clinical test used to evaluate hearing objectively. The aim of this study was to optimise weighted averaging for both residual noise reduction and also for objective ABR detection using the Fmp statistical test. Analyses were performed using no-stimulus EEG background activity recorded from 15 participants and simulated “response present” data (4,602 ensembles in total). Different approaches for estimating the variance of the noise within each block were compared, as was the effect of the number of recording epochs in each block when calculating and applying the weights. The “VAR Whole Block” method was found to be more effective than the “VAR MP” method at estimating the noise level, especially for smaller block sizes (2–10 epochs). Caution should be exerted when selecting recording parameters for use with weighted averaging as an inflation in the “response absent” Fmp statistic was observed using small block sizes (relative to unweighted averaging); this may be due to a bias in the Fmp statistic observed as a result of the combined effects of the finite Fmp analysis window length and the high-pass filter setting. Optimised weighted averaging was effective in reducing the mean residual noise level in the averaged waveform, leading to improved ABR detection. Further work is required to optimise the Fmp analysis window length, recording settings, and weighted averaging parameters in combination, using a large clinical dataset.

A self-attention multi-scale convolutional neural network method for SAR image despeckling

ABSTRACT The speckle noise found in synthetic aperture radar (SAR) images severely affects the efficiency of image interpretation, retrieval and other applications. Thus, effective methods for despeckling SAR image are required. The traditional methods for SAR image despeckling fail to balance in terms of the relationship between the intensity of speckle noise filtering and the retention of texture details. Deep learning based SAR image despeckling methods have been shown to have the potential to achieve this balance. Therefore, this study proposes a self-attention multi-scale convolution neural network (SAMSCNN) method for SAR image despeckling. The advantage of the SAMSCNN method is that it considers both multi-scale feature extraction and channel attention mechanisms for multi-scale fused features. In the SAMSCNN method, multi-scale features are extracted from SAR images through convolution layers with different depths. These are concatenated; then, and an attention mechanism is introduced to assign different weights to features of different scales, obtaining multi-scale fused features with weights. Finally, the despeckled SAR image is generated through global residual noise reduction and image structure fine-tuning. The despeckling experiments in this study involved a variety of scenes using simulated and real data. The performance of the proposed model was analysed using quantitative and qualitative evaluation methods and compared to probabilistic patch-based (PPB), SAR block-matching 3-D (SAR-BM3D) and SAR-CNN methods. The experimental results show that the method proposed in this paper improves the objective indexes and shows great advantages in visual effects compared to these classical methods. The method proposed in this study can provide key technical support for the practical application of SAR images.

Image Processing Framework for Performance Enhancement of Low-Light Image Sensors

Night vision devices are used for capturing images under low light conditions ( $ lux), which are significantly different from images under standard illumination. The night vision devices based on image intensifier tube have power and cost advantages over other night vision devices, such as thermal imaging for short-range surveillance. However, these vision devices suffer from scintillation noise, which makes the resulting images noisy under very low light conditions. This paper presents a comprehensive methodology that uses a consolidated image processing module to improve the performance of low-light images acquired using image-intensifier-based night vision devices. The proposed module is a combination of a two-stage image processing framework. First, a spatio-temporal filter (based on wavelet and recursive frame averaging) is used to minimize inter-frame and intra-frame noise. Then, image enhancement is performed followed by residual noise reduction using a local spatial filter before applying image tone adjustment for display mapping. The results of the proposed module demonstrate performance enhancement of night vision devices under controlled simulated light conditions as well as field conditions. Objective and subjective analysis also reveals that the proposed module can effectively reduce noise and significantly improve visual image quality by 20–30%.

Wearable Hearing Device Spectral Enhancement Driven by Non-Negative Sparse Coding-Based Residual Noise Reduction.

This paper proposes a novel technique to improve a spectral statistical filter for speech enhancement, to be applied in wearable hearing devices such as hearing aids. The proposed method is implemented considering a 32-channel uniform polyphase discrete Fourier transform filter bank, for which the overall algorithm processing delay is 8 ms in accordance with the hearing device requirements. The proposed speech enhancement technique, which exploits the concepts of both non-negative sparse coding (NNSC) and spectral statistical filtering, provides an online unified framework to overcome the problem of residual noise in spectral statistical filters under noisy environments. First, the spectral gain attenuator of the statistical Wiener filter is obtained using the a priori signal-to-noise ratio (SNR) estimated through a decision-directed approach. Next, the spectrum estimated using the Wiener spectral gain attenuator is decomposed by applying the NNSC technique to the target speech and residual noise components. These components are used to develop an NNSC-based Wiener spectral gain attenuator to achieve enhanced speech. The performance of the proposed NNSC–Wiener filter was evaluated through a perceptual evaluation of the speech quality scores under various noise conditions with SNRs ranging from -5 to 20 dB. The results indicated that the proposed NNSC–Wiener filter can outperform the conventional Wiener filter and NNSC-based speech enhancement methods at all SNRs.

Medical Image Denoising Based on Biquadratic Polynomial With Minimum Error Constraints and Low-Rank Approximation

To improve the visual quality of noisy medical images acquired by low radiation dose imaging, medical image denoising is highly desirable for clinical disease diagnosis. In this paper, a geometric regularization method is proposed for medical image denoising. To the best of our knowledge, this is the first work aiming at reconstructing surface by minimizing the gradient error and approximation error of the surface to suppress the noise in medical images. The proposed denoising method consists of two stages: one is to output a basic estimate and the other is for the residual noise reduction. Specifically, the method first exploits a biquadratic polynomial surface to generate an initial estimate of the noise-free image. The surface is constructed by dividing its coefficients into two groups. With the reconstruction error constraint, one group is used to minimize the gradient of the surface, and the other is to minimize the approximation accuracy of the surface. Then the residual noise in the initial result is further reduced by using the singular value thresholding mechanism, which exploits the self-similarity of medical images and the intrinsic low-rank property. Unlike the traditional truncated singular value thresholding scheme, the proposed singular value thresholding is derived by optimizing an objective function with a constraint. Experimental results on a real clinical data set demonstrate the effectiveness of the proposed denoising method, especially in detail-preserving. Compared with several widely used denoising methods, our method can achieve a better performance in terms of both quantitative metrics and subjective visual quality.

Bidirectional microphone array with adaptation controlled by voice activity detector based on multiple beamformers

Ambient noise suppression in a reverberant room is usually performed by the microphone array. The adaptive beamforming, whose typical representative is minimum variance distortionless (MVDR) beamformer, is an effective method for noise suppression. However, MVDR beamformer gives poor results in the real room because of its sensitivity to the steering error and the multipath wave propagation. In this paper we propose a noise suppression method based on assumption that the positions of the speakers in the reverberant room are roughly known. Noise reduction is realized by two MVDR beamformers directed toward each of the speakers. Adaptation of the MVDR beamformers are controlled by a speaker activity detector which decision is based on power transfer model of the multiple superdirective beamformers in combined diffuse and coherent noise field. The proposed voice activity detector also provides residual noise reduction. The proposed method and its robustness to steering error were tested on the model of simulated room as well as in real room environment. The improvement of the restored speech signal was evaluated by Signal to Noise Ratio Enhancement (SNRE) and by Perceptual evaluation of speech quality (PESQ) measure.

Reduction of residual noise based on eigencomponent filtering for speech enhancement

In this paper, residual noise of corrupted speech observations is further restrained based on eigencomponent (an eigenvalue and its corresponding eigenvector) filtering. Three relevant algorithms are proposed to obtain the core eigencomponents that deeply affect enhancement quality of speech fragments by joint diagonalization of clean speech and noise covariance matrix. In addition, the generalized inverse matrix transform is introduced to the recovery of enhanced speech signal for the issue of matrix irreversibility after eigencomponents are filtered. Experiment results show that the proposed methods work better than many other methods under various conditions on both noise reduction and speech distortion.

Noise Estimation and Suppression Using Nonlinear Function withA PrioriSpeech Absence Probability in Speech Enhancement

This paper proposes a noise-biased compensation of minimum statistics (MS) method using a nonlinear function anda priorispeech absence probability (SAP) for speech enhancement in highly nonstationary noisy environments. The MS method is a well-known technique for noise power estimation in nonstationary noisy environments; however, it tends to bias noise estimation below that of the true noise level. The proposed method is combined with an adaptive parameter based on a sigmoid function anda prioriSAP for residual noise reduction. Additionally, our method uses an autoparameter to control the trade-off between speech distortion and residual noise. We evaluate the estimation of noise power in highly nonstationary and varying noise environments. The improvement can be confirmed in terms of signal-to-noise ratio (SNR) and the Itakura-Saito Distortion Measure (ISDM).

A new online secondary path modelling for feedforward ANC systems

Adaptive filter-based feedforward active noise control (ANC) systems are generally realised in system identification configuration. The secondary path of such ANC systems contains numerous subsystems, whose transfer function needs to be compensated for the ANC systems to work satisfactorily. Different methods are available for online modelling of secondary path and realising ANC system using FxLMS algorithm. Simulation of a few such kinds of existing systems is done using white Gaussian noise (WGN) as primary noise and comparing their performances. A new ANC system with online secondary path modelling is proposed and analysed. Simulation study revealed that the proposed ANC system shows superior performance compared to the existing systems in terms of average power reduction of residual noise and converging time when WGN as primary signal.

Performance comparison of interferometer topologies for carrier-envelope phase detection

Performance and noise immunity of different interferometer set-ups for carrier-envelope phase detection are compared. The frequently used Mach–Zehnder interferometer is found to be easily corrupted by acoustic noise contributions and air streaks, whereas a quasi-common-path variant of the f-to-2f interferometer exhibits a 40% reduction of residual noise. This comparative analysis also provides deeper insight into additional mechanisms that are currently limiting the performance of carrier-envelope phase stabilization schemes.

Improved spectral subtraction utilizing iterative processing

AbstractIn this paper, the authors propose a new noise suppression technique that executes iterative processing and sets parameters that are suited for that processing in the spectral subtraction method, which is a noise reduction technique for noise‐added speech. Iterative processing is a technique in which speech enhancement processing is executed by considering the estimated speech that is obtained when noise reduction processing is executed once as the input signal again so that a reduction of the residual noise is anticipated. A further reduction of residual noise for which speech degradation is controlled can be achieved by adjusting the parameters for each iteration. The authors also simultaneously propose a technique for maintaining the real‐time nature of spectral subtraction when the proposed technique is executed. They use actual speech to which white noise, automobile noise, and crowd babble noise is added to compare the characteristics of the two proposed methods with the conventional spectral subtraction method and its improvements. The authors verified according to objective and subjective evaluations that each proposed technique showed superior results in all noisy environments. © 2006 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 90(4): 39 –51, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.20242

Residual noise reduction in sign algorithm

The residual noise of the sign algorithm can be reduced by adding a uniformly distributed noise (dithering) to the sign algorithm input signal. In this paper, a simple equation for the residual noise reduction factor (RNRF) is derived for the case in which the noise in the sign algorithm input signal has a two-level discrete amplitude distribution. It is demonstrated that the RNRF is proportional to the ratio between the amplitude of the input noise and the adaptation step.

Spectral subtraction based on phonetic dependency and masking effects

A speech state-dependent spectral subtraction method to regulate the blind spectral subtraction for improved enhancement is proposed. In this method, a modified subtraction rule is applied over the speech selectively contingent to the speech state being voiced or unvoiced, in an effort to incorporate the acoustic characteristics of phonemes. The aim is to remedy the subtraction induced signal distortion attained by two state-dependent procedures: spectrum sharpening and minimum spectral bound. In order to remove the residual noise, the proposed method employs a procedure utilising the masking effect. The proposed spectral subtraction, including state-dependent subtraction and residual noise reduction using the masking threshold, shows effectiveness in terms of the compensation of spectral distortion in the unvoiced region as well as residual noise reduction.

Telecommunications speech signal improvement by reduction of residual noise

A telecommunications network service overcomes the annoying effects of transmitted noise by signal processing which filters out the noise using a model-based iterative signal estimator. The estimator is provided with a current estimate of the noise power spectral density, using signal frame samples determined by a voice activity detector to be noise-only frames. The signal estimator makes intra-frame iterations of the current frame while using smoothing across LSP parameters of adjacent frames, recent past frames, and up to two contiguous future frames. Non-stationary noise created by the iterative filtering is further reduced in one or more post-filtering stages that use knowledge of the nature of the low level non-stationary noise events.