Noise Enhancement Research Articles

Self-attending RNN for Speech Enhancement to Improve Cross-corpus Generalization.

Deep neural networks (DNNs) represent the mainstream methodology for supervised speech enhancement, primarily due to their capability to model complex functions using hierarchical representations. However, a recent study revealed that DNNs trained on a single corpus fail to generalize to untrained corpora, especially in low signal-to-noise ratio (SNR) conditions. Developing a noise, speaker, and corpus independent speech enhancement algorithm is essential for real-world applications. In this study, we propose a self-attending recurrent neural network (SARNN) for time-domain speech enhancement to improve cross-corpus generalization. SARNN comprises of recurrent neural networks (RNNs) augmented with self-attention blocks and feedforward blocks. We evaluate SARNN on different corpora with nonstationary noises in low SNR conditions. Experimental results demonstrate that SARNN substantially outperforms competitive approaches to time-domain speech enhancement, such as RNNs and dual-path SARNNs. Additionally, we report an important finding that the two popular approaches to speech enhancement: complex spectral mapping and time-domain enhancement, obtain similar results for RNN and SARNN with large-scale training. We also provide a challenging subset of the test set used in this study for evaluating future algorithms and facilitating direct comparisons.

A Computer-Aided Diagnoses Program for Leukemia Detection Using Blood Samples

During the last decades, the number of people given a diagnosis of leukemia had increased significantly. This article aims to automate detection from peripheral blood based on artificial vision methods for more speed and accuracy. The proposed method is applied to data sets from different sources to achieve diversity, which was not pursued before. This required implementation of noise removal, enhancing, and segmentation techniques on the images by applying linear scaling filter, global-local contrast enhancement, and marker-controlled watershed and K-means clustering for each step, respectively. One hundred three features of texture, shape, and color were extracted from cell, nucleus, and cytoplasm subimages, followed by a reduction to 50 most relevant features. The support vector machine model was trained seeking a suitable accuracy to be used in practical experiments. Here, we achieved an accuracy of 91.0256% by implementing the steps previously mentioned in a MATLAB executable program. The software detected leukemia in 5 of 6 images from a new data set, which indicates the usefulness of this approach despite the variation in images.

AUTOMATIC SELECTION OF XRF SPECTRAL FEATURE VARIABLES FOR SOIL HEAVY METAL BASED ON FiPLS AND BiPLS, 52-59.

When using X-ray fluorescence (XRF) spectrometry to detect heavymetal content in soil, there are several factors that affect the range ofspectral variable resolutions such as characteristic peak broadeningof heavy metal, characteristic peak overlapping of heavy metal,instrumental noise, and potential absorption enhancement effect ofelements.

Unsupervised Spectrum Anomaly Detection Method for Unauthorized Bands

With the rapid development of wireless communication, spectrum plays increasingly important role in both military and civilian fields. Spectrum anomaly detection aims at detecting emerging anomaly signals and spectrum usage behavior in the environment, which is indispensable to secure safety and improve spectrum efficiency. However, spectrum anomaly detection faces many difficulties, especially for unauthorized frequency bands. In unauthorized bands, the composition of spectrum is complex and the anomaly usage patterns are unknown in prior. In this paper, a Variational Autoencoder- (VAE-) based method is proposed for spectrum anomaly detection in unauthorized bands. First of all, we theoretically prove that the anomalies in unauthorized bands will introduce Background Noise Enhancement (BNE) effect and Anomaly Signal Disappearance (ASD) effects after VAE reconstruction. Then, we introduce a novel anomaly metric termed as percentile (PER) score, which focuses on capturing the distribution variation of reconstruction error caused by ASD and BNE. In order to verify the effectiveness of our method, we developed an ISM Anomaly Detection (IAD) dataset. The proposed PER score achieves superior performance against different type of anomalies. For QPSK type anomaly, our method increases the recall rate from 80% to 93% while keeping a false alarm rate of 5%. The proposed method is beneficial to broadband spectrum sensing and massive spectrum data processing. The code will be released at git@github.com :QXSLAB/vae_ism_ano.git.

Research on various image processing techniques

Digital image processing deals with the manipulation of digital images through a digital computer. It is a subfield of signals and systems but focuses particularly on images. DIP focuses on developing a computer system that can perform processing on an image. The input of that system is a digital image and the system process that image using efficient algorithms and gives an image as an output. The most common example is Adobe Photoshop. It is one of the widely used applications for processing digital images. The image processing techniques play a vital role in image Acquisition, image pre-processing, Clustering, Segmentation, and Classification techniques with different kinds of images such as Fruits, Medical, Vehicle, and Digital text images, etc. In this study, the various images remove unwanted noise and performance enhancement techniques such as contrast limited adaptive histogram equalization.

Noise Reduction, Enhancement and Classification for Sonar Images

Ultrasound imaging has some problems with image properties output. These affects the specialist decision. Ultrasound noise type is the speckle noise which has a grainy pattern depending on the signal. There are two parts of this study. The first part is the enhancing of images with adaptive Weiner, Lee, Gamma and Frost filters with 3x3, 5x5, and 7x7 sliding windows. The evaluated process was achieved using signal to noise ratio (SNR), peak signal to noise ratio (PSNR), mean square error (MSE), and maximum difference (MD) criteria. The second part consists of simulating noise in a standard image (Lina image) by adding different percentage of speckle noise from 0.01 to 0.06. The supervised classification based minimum distance method is used to evaluate the results depending on selecting four blocks located at different places on the image. Speckle noise was added with different percentage from 0.01 to 0.06 to calculate the coherent noise within the image. The coherent noise was concluded from the slope of the standard deviation with the mean for each noise. The results showed that the additive noise increased with the slide window size, while multiplicative noise did not change with the sliding window nor with increasing noise ratio. Wiener filter has the best results in enhancing the noise.

BRAIN TUMOR CLASSIFICATION USING DEEP LEARNING

The early detection of the tumor plays an important role in the recovery of the patient. In our proposed model, we have collected MRI scans as it helps with the information about the blood supply inside the brain. Thus, for the recognition of anomaly, for examining the increasing of the ailment, and for the diagnosis, we prepared a data set consisting of various MRI images. We then focused on removing unwanted noise and image enhancement. The image characteristics can be enhanced by using image preprocessing techniques. The image enhancement depends upon different factors like computational time, computational cost, quality of the uncorrupted image, and the techniques used for noise elimination. We have made use of various filters for the image pre-processing. In our next step, image segmentation, an image is divided into several regions. We have implemented different types of segmentation techniques including active contours snakes, fuzzy C means, and regionderived triple thresholding. We have further implemented two hybrid segmentation models and used computer-aided detection techniques. Post-processing of the data is done using artificial bee colony optimization and watershed filtering and extraction. We then classify two images into tumor and non-tumor category using the VGG-16 CNN model. The features of the segmented images were further classified into various types of tumors, including Glioma tumor, Meningioma tumor, Pituitary tumor, and no tumor using one-hot encoding. This approach was further validated using synthetic and real MR image dataset from Kaggle (name of data set), to detect and classify different types of tumor.

Multi-User Scheduling for 6G V2X Ultra-Massive MIMO System.

6G vehicle-to-everything (V2X) communication will be combined with vehicle automatic driving technology and play an important role in automatic driving. However, in 6G V2X systems, vehicle users have the characteristics of high-speed movement. Therefore, how to provide stable and reliable wireless link quality and improve channel gain has become a problem that must be solved. To solve this problem, a new multi-user scheduling algorithm based on block diagonalization (BD) precoding for 6G ultra-massive multiple-input multiple-output (MIMO) systems is proposed in this paper. The algorithm takes advantage of the sensitive nature of BD precoding to channel correlation, uses the Pearson coefficient after matrix vectorization to measure the channel correlation between users, defines the scheduling factor to measure the channel quality according to the user noise enhancement factor, and jointly considers the influence of the correlation between user channels and channel quality, ensuring the selection of high-quality channels while minimizing channel correlation. Simulation results show that compared with the multi-user scheduling algorithm based on subspace correlation, condition number, and geometric angle, the proposed algorithm can obtain higher user channel gain, effectively reduce the system bit error rate, and can be applied to 6G V2X communication.

Stochastic resonance of an asymmetric tristable system driven by cross-correlated Ornstein–Uhlenbeck noise

Stochastic resonance of an asymmetric tristable system driven by a periodic forcing and cross-correlated Ornstein–Uhlenbeck (O–U) noise is investigated. An approximate Fokker-Planck Equation for an asymmetric tristable system driven by a periodic forcing and cross-correlated O–U noise is derived in the adiabatic limit by using the unified colored-noise approximation method and the stochastic equivalent rules. Afterwards, the expression for spectral amplification (SA), the stationary probability density function (PDF) and the mean first passage time (MFPT) are derived. By considering the influence of asymmetric potential function and cross-correlated noise, the phenomenon of noise enhancing stability (NES) and stochastic resonance (SR) can be found. It can also be found that the response of the tristable system can be effectively improved by properly selecting the relevant noise intensity and the asymmetric constant. This proves the excellent performance of the system, and provides a good theoretical support for practical engineering application.

Near-Infrared Hyperspectral Imaging as a Monitoring Tool for On-Demand Manufacturing of Inkjet-Printed Formulations

This study evaluates the potential use of near-infrared hyperspectral imaging (NIR-HSI) for quantitative determination of the drug amount in inkjet-printed dosage forms. We chose metformin hydrochloride as a model active pharmaceutical ingredient (API) and printed it onto gelatin films using a piezoelectric inkjet printing system. An industry-ready NIR-HSI sensor combined with a motorized movable linear stage was applied for spectral acquisition. Initial API-substrate screening revealed best printing results for gelatin films with TiO2 filling. For calibration of the NIR-HSI system, escalating drug doses were printed on the substrate. After spectral pre-treatments, including standard normal variate (SNV) and Savitzky-Golay filtering for noise reduction and enhancement of spectral features, principal component analysis (PCA) and partial least squares (PLS) regression were applied to create predictive models for the quantification of independent printed metformin hydrochloride samples. It could be shown that the concentration distribution maps provided by the developed HSI models were capable of clustering and predicting the drug dose in the formulations. HSI model prediction showed significant better correlation to the reference (HPLC) compared to on-board monitoring of dispensed volume of the printer. Overall, the results emphasize the capability of NIR-HSI as a fast and non-destructive method for the quantification and quality control of the deposited API in drug-printing applications.Graphical abstract

Two-time correlation functions beyond quantum regression theorem: effect of external noise

We present the results of a study of the two-time correlation functions of a dichotomously driven two-level system in contact with a thermal bath by using corrections beyond the quantum regression theorem. In the strong system-environment coupling regime, it is found that the noise parameters could be tuned to control the magnitude of corrections at low environmental temperatures. The motional averaging and narrowing effect of the external noise was observed on the absorption and emission spectra of the two-state system. Furthermore, effects similar to the destruction of tunneling and noise enhancement of transport are observed in the dynamics of the two-time correlations.

An Artificial Intelligence Framework for Plant Leaf Disease Detection and Classification Using AMBF with GKFCM and GLCM

As of 2020, the total area planted with crops in India overtook 125.78 million hectares. India is the second biggest organic product maker in the world. Thus, an Indian economy greatly depends on farming products. Nowadays, farmers suffer a drop in production due to a lot of diseases and pests. Thus, to overcome this problem, this article presents the artificial intelligence based deep learning approach for plant disease classification. Initially, the adaptive mean bilateral filter (AMBF) for noise removal and enhancement operations. Then, Gaussian kernel fuzzy C-means (GKFCM) approach is used to segment the effected disease regions. The optimal features from color, texture and shape features are extracted by using GLCM. Finally, Deep learning convolutional neural network (DLCNN) is used for the classification of five class diseases. The segmentation and classification performance of proposed method outperforms as compared with the state of art approaches.

150 m/500 Mbps Underwater Wireless Optical Communication Enabled by Sensitive Detection and the Combination of Receiver-Side Partial Response Shaping and TCM Technology

In this paper, a joint digital signal processing (DSP) scheme including partial response shaping and trellis coded modulation (TCM) technology is proposed to mitigate the bandwidth limitation of the long-reach underwater wireless optical communication (UWOC) system. The similarity of spectral responses between the partial response filter and the communication system is utilized to suppress the enhancement of noise and distortion induced by the full-response feed-forward equalizer. A random interleaver and a precoder are applied to whiten the noise and distortion. In a laboratory experiment, the minimal required received optical power of -29.8 dBm (at a BER of 3.80 ×10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> - 3</sup> ) is realized at a data rate of 500 Mbps. In a standard swimming pool, the proposed system successfully achieves a data rate of 500 Mbps with a transmission distance up to 150 m, at which the total attenuation is measured to be 53.47 dB. Such a data rate, enabled by the proposed joint DSP scheme, is 8.0%, 9.2%, and 14.2% higher than those achieved by solely using partial response shaping scheme, TCM scheme, and least square equalization scheme, respectively. To the best of our knowledge, this is the first time to employ the combination of partial response shaping and TCM technology in UWOC for high-speed long-reach transmission.

A new approach for designing and implementing ADF equalization for 5G frequency selective channel based on two operating phases of LS and RLS algorithms

In this paper, a new approach is proposed for implementing an adaptive decision feedback equalizer (ADFE) for the 5G channel. The proposed equalizer works in two phases. In the first phase, a least-squares (LS) algorithm with a variable-length training sequence is used to estimate the coefficients of the channel and the equalizer. In the second phase, the recursive least-squares algorithm estimates the channel and adapts the equalizer, jointly. According to the channel quality, a variable-length training sequence is used to estimate the channel vector and the coefficients of the equalizer. The feed-forward equalizer (FFE) compensates the effects of the transmitting filter and the channel filter. No matched filter is used in the receiver. The noise samples at the input of the proposed FFE are independent. The noise enhancement of the proposed FFE is less than the noise enhancement of its corresponding one in the conventional ADFE. The overall filtering response (OFR) from the input of the transmitting filter to the output of the FFE is calculated and used to estimate the coefficients of the feedback equalizer (FBE). The channel model, the FFE coefficients, the OFR vector, and the FBE coefficients are continuously updated every symbol period. Using a variable training sequence increases the bandwidth efficiency of the transmitted signal. Simulation results and real-time implementation measurements show that the convergence time and the steady-state error at the output of the proposed equalizer are smaller than their corresponding values in the conventional ADFE.

Performance improvement of data transmission using a hybrid underwater and terrestrial system

AbstractIn this paper, we investigate and enhance the performance of data transmission using a hybrid underwater and terrestrial system with low‐complexity linear equalization. Due to the low speed, salinity, water depth, pH degree, and water temperature variations, underwater acoustic communication has become one of the most challenging technologies in the world. The most common linear equalizers are the Linear Zero Forcing (LZF), and Linear Minimum Mean Square Error (LMMSE) equalizers. The LZF equalizer suffers from the noise enhancement problem, while the LMMSE equalizer requires the value of the operating Signal‐to‐Noise Ratio (SNR) to work, correctly, which increases the computational complexity. In addition, this paper presents a low‐complexity equalization scheme to overcome the drawbacks associated with the LZF, and LMMSE equalizers. The proposed equalizer is called the Joint Low‐Complexity Regularized LZF equalizer, which depends on a fixed regularization parameter to minimize the noise enhancement phenomenon caused by the LZF equalizer. The proposed equalizer does not need the estimation of the SNR, and the computational complexity is reduced due to the utilization of the banded‐matrix approximation. The proposed equalizer is implemented based on the Single‐Input‐Single‐Output configuration for Orthogonal Frequency Division Multiplexing (OFDM) using the Discrete Cosine Transform. The obtained simulation results show that the proposed equalizer outperforms different equalizers for the same channel conditions.

Research on Noise Reduction and Enhancement Algorithm of Girth Weld Image

Research on Noise Reduction and Enhancement Algorithm of Girth Weld Image

Role of Promaxon-D in Controlling Tribological Performance of Cu-Free Brake Pads

Copper appears to be an indispensable ingredient of non-asbestos organic (NAO) brake friction materials owing to its multifunctionality. However, recent findings proved it as a threat to aquatic life and efforts were initiated worldwide over the last few years to find a substitute for it. Recently, the authors’ group had reported on the potential of particulate Promaxon-D (hydrated calcium silicate), a porous material for enhancement of noise–vibration (NV) performance of Cu-free brake pads. It was felt necessary to investigate how Promaxon-D (PD) contributes to controlling tribo-performance of brake pads since this aspect is most important for the brake friction materials. A series of five types of brake pads was developed by keeping parent composition fixed and varying the theme ingredient PD content from 0–20 wt.%. The tribo-performance was evaluated on a full-scale brake inertia dynamometer as per the JASO C406 test standards. The increase in PD contents led to a reduction in density, hardness, and thermal conductivity while an increase in porosity and compressibility. With the increase in PD contents, almost all tribo properties such as average μ, fade ratio, and recovery ratio got enhanced, sensitivity of µ for pressure, speed, and temperature lowered but at the cost of wear resistance. Finally, the worn surfaces of pads and discs were observed under a scanning electron microscope equipped with energy dispersive X-ray (SEM-EDX) to find out the presence of various elements at the contact zone and to monitor specific topographical events responsible for wear mechanism.

Research on Noise Reduction and Enhancement Algorithm of Girth Weld Image

In order to eliminate the salt pepper and Gaussian mixed noise in X-ray weld image, the extreme value characteristics of salt and pepper noise are used to separate the mixed noise, and the non local mean filtering algorithm is used to denoise it. Because the smoothness of the exponential weighted kernel function is too large, it is easy to cause the image details fuzzy, so the cosine coefficient based on the function is adopted. An improved non local mean image denoising algorithm is designed by using weighted Gaussian kernel function. The experimental results show that the new algorithm reduces the noise and retains the details of the original image, and the peak signal-to-noise ratio is increased by 1.5 dB. An adaptive salt and pepper noise elimination algorithm is proposed, which can automatically adjust the filtering window to identify the noise probability. Firstly, the median filter is applied to the image, and the filtering results are compared with the pre filtering results to get the noise points. Then the weighted average of the middle three groups of data under each filtering window is used to estimate the image noise probability. Before filtering, the obvious noise points are removed by threshold method, and then the central pixel is estimated by the reciprocal square of the distance from the center pixel of the window. Finally, according to Takagi Sugeno (T-S) fuzzy rules, the output estimates of different models are fused by using noise probability. Experimental results show that the algorithm has the ability of automatic noise estimation and adaptive window adjustment. After filtering, the standard mean square deviation can be reduced by more than 20%, and the speed can be increased more than twice. In the enhancement part, a nonlinear image enhancement method is proposed, which can adjust the parameters adaptively and enhance the weld area automatically instead of the background area. The enhancement effect achieves the best personal visual effect. Compared with the traditional method, the enhancement effect is better and more in line with the needs of industrial field.

HPWO-LS-based deep learning approach with S-ROA-optimized optic cup segmentation for fundus image classification

Recently, automated retinal image processing has been considered a competitive field of research due to the low-accuracy results, complexity, and unacceptable outcomes associated with it. In this article, we proposed a novel approach for the classification of fundus images from different kinds of fundus disorders. The original images are preprocessed in terms of noise and contrast enhancement by using the contrast limited adaptive histogram equalization method. The optic cup segmentation from the fundus images is effectively handled via the search and rescue optimization algorithm. After that, the color, texture, and shape-based gray-level co-occurrence matrix features are extracted. The hybrid particle swarm optimization with local search strategy improves the DNN parameter and the newly developed method is named as optimal DNN. The optimal DNN is used to classify whether the image is diabetic retinopathy, glaucoma, or age-related macular degeneration. Experimentally, different kinds of datasets such as STARE, Drishti, and RIM-One datasets with performance measure are validated. Finally, the proposed approaches demonstrate higher classification performances in terms of accuracy, specificity, sensitivity, precision, recall, and f-measure.

Enhancement by postfiltering for speech and audio coding in ad hoc sensor networks

Enhancement algorithms for wireless acoustic sensor networks (WASNs) are indispensable with the increasing availability and usage of connected devices with microphones. Conventional spatial filtering approaches for enhancement in WASNs approximate quantization noise with an additive Gaussian distribution, which limits performance due to the non-linear nature of quantization noise at lower bitrates. This work proposes a postfilter for enhancement based on Bayesian statistics to obtain a multidevice signal estimate, which explicitly models the quantization noise. The experiments using perceptual signal-to-noise ratio, perceptual evaluation of speech quality, and MUSHRA (multistimulus with hidden reference and anchors) scores demonstrate that the proposed postfilter can be used to enhance signal quality in ad hoc sensor networks.