Noise Level Estimation Research Articles

Adaptive Real-Time Wavelet Denoising Architecture Based on Feedback Control Loop

The discrete wavelet transform is commonly used as a denoising step for many applications, like biomedical applications which are usually suffering from low SNR of the recorded signal. However, the choice of appropriate threshold value for DWT coefficients plays significant role in reconstructing the denoised signal. This paper presents a design of real-time wavelet denoising architecture which is suitable for wide range of real-time denoising applications. In this design, an adaptive thresholding approach based on feedback control loop is proposed to make the architecture more applicable for real-time wavelet denoising. This thresholding method considers a noise level estimator module based on first detail coefficients level 𝑑1 to calculate the unknown standard deviation of background noise. The proposed architecture is developed using MATLAB to simulate the suggested denoising method. The performance of the proposed denoising method is studied in terms of integral gain 𝐺 of feedback control and window size 𝑀 with respect to the improvement in SNR and settling time. The results imply that the proposed denoising architecture is suitable for real-time denoising applications with acceptable improvement in SNR approximately 8 dB.

Deep neural network based robust computer-aided cataract diagnosis system using fundus retinal images

Cataract is the cloudiness present in the eye lens due to denaturation of active protein cells. Cataract affects the quality-of-life and thereby troubling the daily routine activities. Early diagnosis and treatment may reduce the vision loss and delays the cataract progression. To diagnose large-screen population, the computer-aided cataract diagnosis (CACD) system using fundus retinal (FR) images is required. In this paper, a CACD system using FR images is proposed to achieve better diagnostic accuracy. It is perceived that the performance of existing CACD systems is poor against noisy input FR images. However, the distortion such as noise is unavoidable in input images due to complex processes involved in the image acquisition. Hence, it is required to consider the effect of noise in the design of CACD systems. So the proposed CACD system includes this issue in the design and provides the robust performance. In the presented CACD system, the features are extracted using combined feature extraction (CFE) technique using two independently fine-tuned deep convolutional neural networks. The noise level estimation (NLE)-based classification is adopted in the classification stage. In NLE-based classification, a set of multi-class support vector machine (SVM) classifiers, which are trained independently at noise levels from 0 to 25 are considered. Finally, the features extracted using CFE are then mapped to a specific multi-class SVM classifier based on noise level present in an input FR image. From the experimental results, it is observed that the proposed system exhibits superior performance than existing CACD systems under noisy conditions.

Gaussian noise level estimation for color image denoising.

Noise level is an important parameter in many visual applications, especially in image denoising. How to accurately estimate the noise level from a noisy image is a challenging problem. However, for color image denoising, it is not that the more accurate the noise level is, the better the denoising performance is, but that the noise level higher than the true noise can achieve a better denoising result. For better denoising, we propose a statistical iterative method based on low-rank image patches. We select the low-rank patches in the image and calculate the eigenvalues of the covariance matrix of these patches. Unlike the existing methods that take the smallest eigenvalue as the estimated noise level, the proposed method analyzes the relationship between the median value and the mean value of the eigenvalue according to the statistical property and selects an appropriate number of eigenvalues to average as the estimated noise level. Extensive experiments are conducted, demonstrating that our estimated noise level reaches the highest value of all comparison methods. And the denoising results on color images of our method outperform all the state-of-the-art methods and the true noise level.

Transform and statistical parameter-based image noise level prediction

Noise level estimation in an image is important and useful in many image processing algorithms such as image denoising, image segmentation, and image compression. Accurately estimating the noise level without prior knowledge of the image is the major challenge of today’s research. We present an improved patch-based fast noise level estimation using DCT and standard deviation method for fast and reliable noise level estimation and the result is compared with the available state-of-art methods. Experimental result shows the proposed method provides greater accuracy, the stability and also the proposed method is an average of six times faster than that of the state- of – art methods for noise level estimation.

Image Denoising Using a Novel Deep Generative Network with Multiple Target Images and Adaptive Termination Condition

Image denoising, a classic ill-posed problem, aims to recover a latent image from a noisy measurement. Over the past few decades, a considerable number of denoising methods have been studied extensively. Among these methods, supervised deep convolutional networks have garnered increasing attention, and their superior performance is attributed to their capability to learn realistic image priors from a large amount of paired noisy and clean images. However, if the image to be denoised is significantly different from the training images, it could lead to inferior results, and the networks may even produce hallucinations by using inappropriate image priors to handle an unseen noisy image. Recently, deep image prior (DIP) was proposed, and it overcame this drawback to some extent. The structure of the DIP generator network is capable of capturing the low-level statistics of a natural image using an unsupervised method with no training images other than the image itself. Compared with a supervised denoising model, the unsupervised DIP is more flexible when processing image content that must be denoised. Nevertheless, the denoising performance of DIP is usually inferior to the current supervised learning-based methods using deep convolutional networks, and it is susceptible to the over-fitting problem. To solve these problems, we propose a novel deep generative network with multiple target images and an adaptive termination condition. Specifically, we utilized mainstream denoising methods to generate two clear target images to be used with the original noisy image, enabling better guidance during the convergence process and improving the convergence speed. Moreover, we adopted the noise level estimation (NLE) technique to set a more reasonable adaptive termination condition, which can effectively solve the problem of over-fitting. Extensive experiments demonstrated that, according to the denoising results, the proposed approach significantly outperforms the original DIP method in tests on different databases. Specifically, the average peak signal-to-noise ratio (PSNR) performance of our proposed method on four databases at different noise levels is increased by 1.90 to 4.86 dB compared to the original DIP method. Moreover, our method achieves superior performance against state-of-the-art methods in terms of popular metrics, which include the structural similarity index (SSIM) and feature similarity index measurement (FSIM). Thus, the proposed method lays a good foundation for subsequent image processing tasks, such as target detection and super-resolution.

CHARACTERISTICS OF MICROSEISMS AND ACOUSTIC NOISES IN THE TRANSPORT POLYGON CONDITIONS

The work is related to numerical estimation and comparative analysis of microseismic and acoustic noise levels in transport polygon conditions. The aim of the work is to study and further define the difference between the signal and noise to improve the ability to detect poorly distinguishable events, as well as to study the propagation features of the interrelated seismic and acoustic wave fields. It is related to the fact that wave processes generated by many natural and anthropogenic sources are conjugate nature. In particular, it is related to simultaneous propagation of seismic waves in the ground and acoustic waves in the atmosphere. It determines expediency of taking into account simultaneously parameters of both types of waves in the tasks of geophysical monitoring of the environment, as well as in solving some applied problems of seismics and acoustics.

Noise level estimation for effective blind despeckling of medical ultrasound images

Despeckling of medical ultrasound images is an essential pre-processing step in automated diagnosis systems. The performance of certain image despeckling methods is dependent on the noise level of an input image. In the design of despeckling algorithms, the noise level is assumed to be known. Therefore, it is essential to estimate the noise level from input ultrasound image for effective blind despeckling. In this paper, a novel noise level estimation (NLE) technique is proposed to estimate the speckle (signal-dependent) noise level from the noisy ultrasound images. The presented NLE technique uses noise level aware features obtained from the high-pass filtered noisy image. The extracted features are then used to train the support vector regression (SVR) model. The proposed feature extraction technique followed by the SVR model results in accurate NLE. In order to ensure the effectiveness, the proposed NLE technique is incorporated and validated with the NLE-dependent state-of-the-art despeckling methods. The combination of state-of-the-art despeckling methods with the proposed NLE technique shows superior despeckling performance when compared to existing NLE techniques. In this work, block matching based despeckling method is recommended in combination with the proposed NLE technique for better despeckling performance in the ultrasound images. The experimental results demonstrate that the proposed NLE technique shows better performance than existing NLE techniques in terms of average absolute deviation and execution time.

Variability of traffic noise pollution levels as a function of city size variables

Noise levels measured in 27 cities with different areas (from 0.6 km2 to 59.27 km2) and populations (from approximately 2000 to 70,000 inhabitants) were compared with respect to five different urban characteristics (population, area, total street length, density, and linear density). Comparisons were conducted for both overall city noise levels and noise registered on five types of roads with different functionality using the Categorisation Method. The results showed that four of the five cities’ characteristics presented a significant correlation with the noise levels (all except for density). The calculated correlations were better for noise levels in the different categories than the overall noise values, with higher explained variability on the streets with more traffic. Therefore, the road categorisation method can be used not only to assess the noise variations within cities, but also to better explain the effect of noise on the analysed city characteristics. The results of the calculated relationships enable the estimation of noise levels both currently and in future urban developments of noise values on different types of streets.

A Skewness Fitting Model for Noise Level Estimation and the Applications in Image Denoising

In this paper, a novel noise level estimation technique is presented based on skewness fitting. The difference between exact skewness and observed value on each channel contributes to an objective function. The skewness concentration is helpful to guarantee the reasonability of the model. The optimal solution means an estimation of the noise variance and it is easy to be applied in some parametric denoising model. The accuracy and efficiency of proposed algorithm were verified by the numerical experiments.

Classification of Space Micromotion Targets With Similar Shapes at Low SNR

There are a number of targets in space that are similar in shape, but they pose different threat levels. It is necessary to discriminate these targets for the sake of space security. The unique characteristics of detailed micro-Doppler (MD) signatures can be used to classify micromotion forms, such as spinning, precession, nutation, wobbling, and tumbling. In this letter, we design a novel processing flow for classifying five types of micromotion under low signal-to-noise ratio (SNR) condition. We start from the range map and successively perform noise-level estimation, adaptive filtering, and multithreshold segmentation to obtain an ideal binary mask. The clean range maps are used as the final input to an eight-layer convolutional neural network (CNN) for training, validation, and testing. The experiment results show that our method achieves more than 80% classification accuracy for micromotion in the case of −10 dB, which is higher than the existing popular methods.

Deep noise estimation and removal for real-world noisy images

This paper proposes a joint noise estimation and removal network for real-world noisy images. We observe that the noise variance depends on the image intensity and the variances of neighboring pixels are not similar due to the demosaicing interpolation. Therefore, for noise estimation, we propose to pack each 2 × 2 noisy block into a 4D vector as the input of the noise level estimation network, which makes the noise variances of adjacent pixels to be similar. This strategy not only reduces the computational complexity but also improves the estimation accuracy. For noise removal, we propose to concatenate the estimated noise variance with the noisy image as the input of the noise removal network, which takes advantages of the correlations between noise levels and image intensities. In addition, we propose a RSD (Residual SE D-Conv) block, which explores the spatial and channel correlations of realistic noise. Considering the realistic noisy and clean image pairs are limited for training, we utilize synthetic noisy images to train the network first and then fine tune the network with real images. Experimental results demonstrate that the proposed method outperforms competing methods for both noise estimation and removal tasks.

Hybrid BM3D and PDE filtering for non-parametric single image denoising

The BM3D method achieves excellent denoising performance, but it has artificial effects and bias effects and its performance largely depends on the noise level parameter. To address this, we propose a hybrid BM3D and PDE method for non-parametric single image denoising. First, a non-local Perona–Malik (NLPM) filtering is proposed, and we prove its discontinuity maintaining, mean invariance, convergence, and local continuity. Based on these mathematical properties, an NLPM based noise level estimator (NLPM-NLE) is explored, which involves three steps: preprocessing by NLPM filtering, sample area selection, parameter estimation. And then, we advance a stable-BM3D (SBM3D) method with NLPM filtering to avoid artificial effects and bias effects. Finally, connecting the NLPM-NLE and SBM3D by merging the same part, we develop a non-parametric single image denoising (NPSID) method. Additionally, our proposed BM3D method with NLPM-NLE and the NPSID are compared with other blind denoising methods including PCA + BM3D, WTP + BM3D, and ESM + BM3D on real image denoising. Experiments show that the proposed non-parametric method can automatically and effectively remove noise and preserve details.

Data Preparation Protocol for Low Signal-to-Noise Ratio Fluorine-19 MRI.

Fluorine-19 MRI shows great promise for a wide range of applications including renal imaging, yet the typically low signal-to-noise ratios and sparse signal distribution necessitate a thorough data preparation.This chapter describes a general data preparation workflow for fluorine MRI experiments. The main processing steps are: (1) estimation of noise level, (2) correction of noise-induced bias and (3) background subtraction. The protocol is supplemented by an example script and toolbox available online.This chapter is based upon work from the COST Action PARENCHIMA, a community-driven network funded by the European Cooperation in Science and Technology (COST) program of the European Union, which aims to improve the reproducibility and standardization of renal MRI biomarkers. This analysis protocol chapter is complemented by two separate chapters describing the basic concept and experimental procedure.

Comparison of Road Traffic Noise prediction models: CNOSSOS-EU, Nord2000 and TRANEX.

Road traffic noise is the most pervasive source of ambient outdoor noise pollution in Europe. Traffic noise prediction models vary in parameterisation and therefore may produce different estimates of noise levels depending on the geographical setting in terms of emissions sources and propagation field. This paper compares three such models: the European standard, Common Noise Assessment Methods for the EU Member States (hereafter, CNOSSOS), Nord2000 and Traffic Noise Exposure (TRANEX) model based on the UK methodology, in terms of their source and propagation characteristics. The tools are also compared by analysing estimated noise (LAeq) from CNOSSOS, Nord2000 (2006 version), and TRANEX for more than one hundred test cases (N=111) covering a variety of source and receiver configurations (e.g. varying source to receiver distance). The main aim of this approach was to investigate the potential pattern in differences between models' performance for certain types of configurations. Discrepancies in performance may thus be linked to the differences in parameterisations of the CNOSSOS, Nord2000, and TRANEX (e.g. handling of diffraction, refraction). In most cases, both CNOSSOS and TRANEX reproduced LAeq levels of Nord2000 (2006 version) within three to five dBA (CNOSSOS: 87%, TRANEX: 94%). The differences in LAeq levels of CNOSSOS, compared to Nord2000, can be related to several shortcomings of the existing CNOSSOS algorithms (e.g. ground attenuation, multiple diffractions, and mean ground plane). The analyses show that more research is required in order to improve CNOSSOS for its implementation in the EU. In this context, amendments for CNOSSOS proposed by an EU Working Group hold significant potential. Overall, both CNOSSOS and TRANEX produced similar results, with TRANEX reproducing Nord2000 LAeq values slightly better than the CNOSSOS. The lack of measured noise data highlights one of the significant limitations of this study and needs to be addressed in future work.

Estimation of Self Noise of Geomagnetic and Geoelectric Field Sensors by the Method of Multichannel Measurements

The application of the methods of electromagnetic sounding for geophysical research is considered. The results of synchronous measurements of the Earth’s electric and magnetic fields made by different types of sensors in the frequency range of 0.002–7000 Hz are presented. Various approaches to obtaining the estimate of sensor noise are considered. A technique for quantitative estimation of the self noise level of the sensors under study and the characteristics of the natural field on the basis of two- and three-channel schemes of synchronous measurements is proposed. Comparison of experimental noise estimates with computational data is given. It is shown that the magnetometer based on a magnetomodulation transducer has an unacceptably high level of self noise for using it in applications related to the measurement of natural variable magnetic field.

Exposure to the road traffic noise in an urban complex in Greece: the quantification of healthy life years lost due to noise-induced annoyance and noise-induced sleep disturbances.

In the recent years, the environmental noise is a global issue of great concern. Especially, it is considered to be one of the most deleterious environmental risk factors for the human health and well-being in urban areas. In this study, we focus on the major source of the transportation noise in the modern-day urbanized societies, which is the road traffic noise. The study was performed in the urban complex in Thessaloniki-Neapoli in Greece, and the estimation of the road traffic noise levels was based on the CoRTN (Calculation of Road Traffic Noise) prediction method. In addition, we estimated the EBD (environmental burden of disease), in terms of DALYs (disability-adjusted life years), due to the annoyance and sleep disturbances associated with the exposure to road traffic noise. The estimation of the EBD was based on the strategic noise mapping in Thessaloniki-Neapoli as well as on the guidance document produced by the WHO (World Health Organization) for the quantitative assessment on the humans' health consequences of the environmental noise. Above all, the results revealed the magnitude of the health damage caused by the transportation noise.

Learning an adaptive model for extreme low‐light raw image processing

Low-light images suffer from severe noise and low illumination. In this work, the authors propose an adaptive low-light raw image enhancement network to avoid parameter-handcrafting in current deep learning models and to improve image quality. The proposed method can be divided into two sub-models: brightness prediction and exposure shifting (ES). The former is designed to control the brightness of the resulting image by estimating a guideline exposure time . The latter learns to approximate an exposure-shifting operator ES, converting a low-light image with real exposure time to a noise-free image with guideline exposure time . Additionally, structural similarity loss and image enhancement vector are introduced to promote image quality, and a new campus image data set (CID) for training the proposed model is proposed to overcome the limitations of the existing data sets. In quantitative tests, it is shown that the proposed method has the lowest noise level estimation score compared with the state-of-the-art low-light algorithms, suggesting a superior denoising performance. Furthermore, those tests illustrate that the proposed method is able to adaptively control the global image brightness according to the content of the image scene. Lastly, the potential application in video processing is briefly discussed.

Traffic noise assessment based on mobile measurements

Continuous noise monitoring based on mobile systems could provide a quick feedback to assess the effect of policies implemented by authorities to mitigate noise pollution. The present research verifies that mobile measurements taken along a main street and aggregated in time and space can accurately estimate noise levels at static points. As a consequence, mobile sensors would be suitable to build, and continuously update, noise maps. Furthermore, the experiment computes the optimum aggregation distance of the mobile measurements.To perform the mobile noise measurements, a low-cost noise sensor with an integrated GPS was mounted on a bicycle. One hour worth of measurements was taken along a main avenue with the mobile receiver simultaneously to 6 static measurement points. For the mobile receiver, the LAeq was computed aggregating samples within a radius from 1 m to 100 m around the static measurement points. Then, the error between the aggregated LAeq of the mobile and the static receivers for the same time period was computed. It is observed that the RMSE and the measurement uncertainties decrease as the aggregation distance increases, having a minimum at an aggregation radius of 33 m and reaching a stabilization due to the constant traffic of the studied street.

Introducing an Improved GRACE Global Point-Mass Solution—A Case Study in Antarctica

In the so-called point-mass modeling, surface densities are represented by point masses, providing only an approximated solution of the surface integral for the gravitational potential. Here, we propose a refinement for the point-mass modeling based on Taylor series expansion in which the zeroth-order approximation is equivalent to the point-mass solution. Simulations show that adding higher-order terms neglected in the point-mass modeling reduces the error of inverted mass changes of up to 90% on global and Antarctica scales. The method provides an alternative to the processing of the Level-2 data from the Gravity Recovery and Climate Experiment (GRACE) mission. While the evaluation of the surface densities based on improved point-mass modeling using ITSG-Grace2018 Level-2 data as observations reveals noise level of approximately 5.77 mm, this figure is 5.02, 6.05, and 5.81 mm for Center for Space Research (CSR), Goddard Space Flight Center (GSFC), and Jet Propulsion Laboratory (JPL) mascon solutions, respectively. Statistical tests demonstrate that the four solutions are not significant different (95% confidence) over Antarctica Ice Sheet (AIS), despite the slight differences seen in the noises. Therefore, the estimated noise level for the four solutions indicates the quality of GRACE mass changes over AIS. Overall, AIS shows a mass loss of −7.58 mm/year during 2003–2015 based on the improved point-mass solution, which agrees with the values derived from mascon solutions.

Seeded intervals and noise level estimation in change point detection: a discussion of Fryzlewicz (2020)

In this discussion, we compare the choice of seeded intervals and that of random intervals for change point segmentation from practical, statistical and computational perspectives. Furthermore, we investigate a novel estimator of the noise level, which improves many existing model selection procedures (including the steepest drop to low levels), particularly for challenging frequent change point scenarios with low signal-to-noise ratios.