Noisy Regions Research Articles

Improvement of Criminisi’s Stripe Noise Suppression Method for Side-Scan Sonar Images

In response to the problem of stripe noise significantly reducing the clarity and details of side-scan sonar images due to various factors, the authors of this paper propose an improved Criminisi method for stripe noise suppression. To address the issues encountered in the Criminisi algorithm during the suppression of stripe noise in side-scan sonar images, the following steps are suggested: firstly, introduce dynamic weights in the priority calculation to adaptively adjust the confidence and data term weights based on the current patch’s texture complexity; secondly, utilize the Sobel operator in the data term calculation to capture the image edge information more accurately; and, thirdly, optimize the matching block search process by introducing the Manhattan distance in addition to the Sum of Squared Differences (SSD) criterion to further select the best matching block while transitioning from a global search to a local search. Experimental validation was conducted using simulated stripe noise images, comparing the proposed method with four traditional denoising techniques. The results demonstrate that the denoising effectiveness of the proposed method is superior, effectively restoring texture in noisy regions while preserving texture structure integrity. Ablation experiments validate the effectiveness of the proposed improvements. Denoising experiments on real noisy images show satisfactory results with this method, and combining it with Fourier transform for additional smoothing in certain cases may yield even better results.

Improving microvascular sensitivity of color Doppler using phase mask based flow recycling algorithm


Blood flow sensitivity is a crucial metric for appraising the effectiveness of color Doppler flow imaging (CDFI). Color Doppler velocity maps based on classic autocorrelation techniques are widely used in clinical practice. However, these techniques often produce twinkling artifacts in noisy regions due to the inherent randomness of noise phases. To mitigate artifacts and improve image quality, Power Mask (PoM) technology becomes imperative. Nevertheless, PoM technology unintentionally filters out small flow signals that have similar power and frequency characteristics to noise signals, thereby reducing the imaging system's sensitivity to flow. 
Approach:
To address this issue, a novel Flow Recycling Algorithm (FRA) based on phase anomaly is introduced in this study. This algorithm, excavating small flow signals from noise, aims to enhance the small flow signals with low-velocity by the phase characteristics of the color Doppler flow information. 
Main results: 
Experiments in multi-organ imaging have shown that the FRA-CDFI approach is more effective in suppressing twinkling artifacts in noisy regions, preserving intricate small flow signals, and markedly improving small blood flow sensitivity. This novel approach provides adequate technical support for clinical ultrasound imaging of organs with dense small blood vessels, such as the brain, kidneys, liver, and more. 
Significance: 
As a novel post-processing method, FRA-CDFI holds significant potential for future deployment in clinical high-frame-rate ultrasound imaging devices.

MetaSeg: Content-Aware Meta-Net for Omni-Supervised Semantic Segmentation.

Noisy labels, inevitably existing in pseudo-segmentation labels generated from weak object-level annotations, severely hamper model optimization for semantic segmentation. Previous works often rely on massive handcrafted losses and carefully tuned hyperparameters to resist noise, suffering poor generalization capability and high model complexity. Inspired by recent advances in meta-learning, we argue that rather than struggling to tolerate noise hidden behind clean labels passively, a more feasible solution would be to find out the noisy regions actively, so as to simply ignore them during model optimization. With this in mind, this work presents a novel meta-learning-based semantic segmentation method, MetaSeg, that comprises a primary content-aware meta-net (CAM-Net) to serve as a noise indicator for an arbitrary segmentation model counterpart. Specifically, CAM-Net learns to generate pixel-wise weights to suppress noisy regions with incorrect pseudo-labels while highlighting clean ones by exploiting hybrid strengthened features from image content, providing straightforward and reliable guidance for optimizing the segmentation model. Moreover, to break the barrier of time-consuming training when applying meta-learning to common large segmentation models, we further present a new decoupled training strategy that optimizes different model layers in a divide-and-conquer manner. Extensive experiments on object, medical, remote sensing, and human segmentation show that our method achieves superior performance, approaching that of fully supervised settings, which paves a new promising way for omni-supervised semantic segmentation.

Bayesian reweighting of biomolecular structural ensembles using heterogeneous cryo-EM maps with the cryoENsemble method

Cryogenic electron microscopy (cryo-EM) has emerged as a powerful method for the determination of structures of complex biological molecules. The accurate characterisation of the dynamics of such systems, however, remains a challenge. To address this problem, we introduce cryoENsemble, a method that applies Bayesian reweighting to conformational ensembles derived from molecular dynamics simulations to improve their agreement with cryo-EM data, thus enabling the extraction of dynamics information. We illustrate the use of cryoENsemble to determine the dynamics of the ribosome-bound state of the co-translational chaperone trigger factor (TF). We also show that cryoENsemble can assist with the interpretation of low-resolution, noisy or unaccounted regions of cryo-EM maps. Notably, we are able to link an unaccounted part of the cryo-EM map to the presence of another protein (methionine aminopeptidase, or MetAP), rather than to the dynamics of TF, and model its TF-bound state. Based on these results, we anticipate that cryoENsemble will find use for challenging heterogeneous cryo-EM maps for biomolecular systems encompassing dynamic components.

DSTAN: A Deformable Spatial-temporal Attention Network with Bidirectional Sequence Feature Refinement for Speckle Noise Removal in Thyroid Ultrasound Video.

Thyroid ultrasound video provides significant value for thyroid diseases diagnosis, but the ultrasound imaging process is often affected by the speckle noise, resulting in poor quality of the ultrasound video. Numerous video denoising methods have been proposed to remove noise while preserving texture details. However, existing methods still suffer from the following problems: (1) relevant temporal features in the low-contrast ultrasound video cannot be accurately aligned and effectively aggregated by simple optical flow or motion estimation, resulting in the artifacts and motion blur in the video; (2) fixed receptive field in spatial features integration lacks the flexibility of aggregating features in the global region of interest and is susceptible to interference from irrelevant noisy regions. In this work, we propose a deformable spatial-temporal attention denoising network to remove speckle noise in thyroid ultrasound video. The entire network follows the bidirectional feature propagation mechanism to efficiently exploit the spatial-temporal information of the whole video sequence. In this process, two modules are proposed to address the above problems: (1) a deformable temporal attention module (DTAM) is designed after optical flow pre-alignment to further capture and aggregate relevant temporal features according to the learned offsets between frames, so that inter-frame information can be better exploited even with the imprecise flow estimation under the low contrast of ultrasound video; (2) a deformable spatial attention module (DSAM) is proposed to flexibly integrate spatial features in the global region of interest through the learned intra-frame offsets, so that irrelevant noisy information can be ignored and essential information can be precisely exploited. Finally, all these refined features are rectified and merged through residual convolution blocks to recover the clean video frames. Experimental results on our thyroid ultrasound video (US-V) dataset and the DDTI dataset demonstrate that our proposed method exceeds 1.2 1.3dB on PSNR and has clearer texture detail compared to other state-of-the-art methods. In the meantime, the proposed model can also assist thyroid nodule segmentation methods to achieve more accurate segmentation effect, which provides an important basis for thyroid diagnosis. In the future, the proposed model can be improved and extended to other medical image sequence datasets, including CT and MRI slice denoising. The code and datasets are provided at https://github.com/Meta-MJ/DSTAN .

Enricherator: A Bayesian Method for Inferring Regularized Genome-wide Enrichments from Sequencing Count Data

A pervasive question in biological research studying gene regulation, chromatin structure, or genomics is where, and to what extent, does a signal of interest arise genome-wide? This question is addressed using a variety of methods relying on high-throughput sequencing data as their final output, including ChIP-seq for protein-DNA interactions,1 GapR-seq for measuring supercoiling,2 and HBD-seq or DRIP-seq for R-loop positioning.3,4 Current computational methods to calculate genome-wide enrichment of the signal of interest usually do not properly handle the count-based nature of sequencing data, they often do not make use of the local correlation structure of sequencing data, and they do not apply any regularization of enrichment estimates. This can result in unrealistic estimates of the true underlying biological enrichment of interest, unrealistically low estimates of confidence in point estimates of enrichment (or no estimates of confidence at all), unrealistic gyrations in enrichment estimates at very close (<10 bp) genomic loci due to noise inherent in sequencing data, and in a multiple-hypothesis testing problem during interpretation of genome-wide enrichment estimates. We developed a tool called Enricherator to infer genome-wide enrichments from sequencing count data. Enricherator uses the variational Bayes algorithm to fit a generalized linear model to sequencing count data and to sample from the approximate posterior distribution of enrichment estimates (https://github.com/jwschroeder3/enricherator). Enrichments inferred by Enricherator more precisely identify known binding sites in cases where low coverage between binding sites leads to false-positive peak calls in these noisy regions of the genome; these benefits extend to published datasets.

INCREASING ROBUSTNESS OF I-VECTORS VIA MASKING: A CASE STUDY IN SYNTHETIC SPEECH DETECTION

Ensuring security in speaker recognition systems is crucial. In the past years, it has been demonstrated that spoofing attacks can fool these systems. In order to deal with this issue, spoof speech detection systems have been developed. While these systems have served with a good performance, their effectiveness tends to degrade under noise. Traditional speech enhancement methods are not efficient for improving performance, they even make it worse. In this research paper, performance of the noise mask obtained via a convolutional neural network structure for reducing the noise effects was investigated. The mask is used to suppress noisy regions of spectrograms in order to extract robust i-vectors. The proposed system is tested on the ASVspoof 2015 database with three different noise types and accomplished superior performance compared to the traditional systems. However, there is a loss of performance in noise types that are not encountered during training phase.

Weakly Supervised Semantic Segmentation for Driving Scenes

State-of-the-art techniques in weakly-supervised semantic segmentation (WSSS) using image-level labels exhibit severe performance degradation on driving scene datasets such as Cityscapes. To address this challenge, we develop a new WSSS framework tailored to driving scene datasets. Based on extensive analysis of dataset characteristics, we employ Contrastive Language-Image Pre-training (CLIP) as our baseline to obtain pseudo-masks. However, CLIP introduces two key challenges: (1) pseudo-masks from CLIP lack in representing small object classes, and (2) these masks contain notable noise. We propose solutions for each issue as follows. (1) We devise Global-Local View Training that seamlessly incorporates small-scale patches during model training, thereby enhancing the model's capability to handle small-sized yet critical objects in driving scenes (e.g., traffic light). (2) We introduce Consistency-Aware Region Balancing (CARB), a novel technique that discerns reliable and noisy regions through evaluating the consistency between CLIP masks and segmentation predictions. It prioritizes reliable pixels over noisy pixels via adaptive loss weighting. Notably, the proposed method achieves 51.8\% mIoU on the Cityscapes test dataset, showcasing its potential as a strong WSSS baseline on driving scene datasets. Experimental results on CamVid and WildDash2 demonstrate the effectiveness of our method across diverse datasets, even with small-scale datasets or visually challenging conditions. The code is available at https://github.com/k0u-id/CARB.

Dynamic mask generation based on peak to correlation energy ratio for light reflection and shadow in PIV images

The performance of particle image velocimetry (PIV) measurements directly depends on the identification and removal of noisy regions in the field of view. Regions in the image with shadows and light reflections directly affect the PIV correlation and consequently deteriorate the vector field. This work proposes a new approach for mask generation based on peak correlation energy (PCE) ratio to remove the noisy effects of light reflection and shadow in PIV images. This approach considers that it is possible to relate low and high PCE values to noise and signal, respectively. Based on this, the PCE values are organized in a probability density function and a threshold is automatically estimated to separate the two regions. Furthermore, a mask correction step is proposed in the present method. The efficacy of this method is demonstrated through cases using both synthetic and experimental images, showcasing its potential to improve PIV analysis accuracy.

UCFN[sbnd]Net: Ulcerative colitis evaluation based on fine-grained lesion learner and noise suppression gating

Background and ObjectiveUlcerative colitis (UC) is a chronic disease characterized by recurrent symptoms and significant morbidity. The exact cause of the disease remains unknown. The selection of current treatment options for ulcerative colitis depends on the severity and location of the disease in each patient. Therefore, developing a fully automated endoscopic images for evaluating UC is crucial for guiding treatment plans and facilitating early prevention efforts. MethodsWe propose a network called ulcerative colitis evaluation based on fine-grained lesion learner and noise suppression gating (UCFNNet). UCFNNet contains three novel modules. Firstly, a fine-grained lesion feature learner (FG-LF Learner) is proposed by integrating local features and a Softmax category prediction (SCP) module to improve the feature accuracy in small lesion areas. Subsequently, a graph convolutional feature combiner (GCFC) is developed to connect features across adjacent convolutional layers and to incorporate short connections between input and output, thereby mitigating feature loss during transmission. Thereafter, a noise suppression gating (NS gating) technique is designed by implementing a grid attention mechanism and a feature gating (FG) module to prioritize significant lesion features and suppress irrelevant and noisy regions in the input feature map. ResultsWe evaluate the performance of the proposed network on both privately-collected and publicly-available datasets. The evaluation of UC achieves excellent results on privately-collected dataset, with an accuracy (ACC) of 89.57 %, Matthews correlation coefficient (MCC) of 85.52 %, precision of 89.26 %, recall of 89.48 %, and F1-score of 89.78 %. The results are also impressive on publicly-available dataset, with ACC of 85.47 %, MCC of 80.42 %, precision of 85.62 %, recall of 84.00 %, and F1-score of 84.53 %, surpassing the performance of state-of-the-art techniques. ConclusionOur proposed model introduces three innovative algorithm modules, which outperform the current state-of-the-art methods and achieve high ACC and F1-score. This indicates that our method has superior performance compared to traditional machine learning and existing deep methods, which means that our method has good application prospects. Meanwhile, it has been verified that the proposed model demonstrates good interpretability. The source code is available at github.com/YinLeRenNB/UCFNNet.

Beyond Photometric Consistency: Geometry-Based Occlusion-Aware Unsupervised Light Field Disparity Estimation.

Although learning-based light field disparity estimation has achieved great progress in the most recent years, the performance of unsupervised light field learning is still hindered by occlusions and noises. By analyzing the overall strategy underlying the unsupervised methodology and the light field geometry implied in epipolar plane images (EPIs), we look beyond the photometric consistency assumption, and design an occlusion-aware unsupervised framework to deal with the situations of photometric consistency conflict. Specifically, we present a geometry-based light field occlusion modeling, which predicts a group of visibility masks and occlusion maps, respectively, by forward warping and backward EPI-line tracing. In order to learn better the noise-and occlusion-invariant representations of the light field, we propose two occlusion-aware unsupervised losses: occlusion-aware SSIM and statistics-based EPI loss. Experiment results demonstrate that our method can improve the estimation accuracy of light field depth over the occluded and noisy regions, and preserve the occlusion boundaries better.

Adequate alignment and interaction for cross-modal retrieval

Cross-modal retrieval has attracted widespread attention in many cross-media similarity search applications, especially image-text retrieval in the fields of computer vision and natural language processing. Recently, visual and semantic embedding (VSE) learning has shown promising improvements on image-text retrieval tasks. Most existing VSE models employ two unrelated encoders to extract features, then use complex methods to contextualize and aggregate those features into holistic embeddings. Despite recent advances, existing approaches still suffer from two limitations: 1) without considering intermediate interaction and adequate alignment between different modalities, these models cannot guarantee the discriminative ability of representations; 2) existing feature aggregators are susceptible to certain noisy regions, which may lead to unreasonable pooling coefficients and affect the quality of the final aggregated features. To address these challenges, we propose a novel cross-modal retrieval model containing a well-designed alignment module and a novel multimodal fusion encoder, which aims to learn adequate alignment and interaction on aggregated features for effectively bridging the modality gap. Experiments on Microsoft COCO and Flickr30k datasets demonstrates the superiority of our model over the state-of-the-art methods.

Crosswalk Detection from Satellite Imagery for Pedestrian Network Completion

Pedestrian network creation and completion for various tasks in transportation planning engineering generally requires laborious manual labeling of sidewalks and crosswalks. In this study, we employed the deep learning-based object detection model YOLO v5 for detecting crosswalks using Google Maps satellite imagery data of four North American cities. We also tested for two additional scenarios involving two image-processing techniques to exploit the special properties of crosswalks. Observations showed that the original images performed better than with binary thresholding and nearly the same even when noisy regions outside the right-of-way were removed. We proposed an algorithm to assign the detected crosswalks to intersection legs. We demonstrated the effectiveness of this technique for Washington, D.C. and Los Angeles, CA, showing classification accuracy rates of 71% and 89%, respectively. We also showed the influence of increasing distance threshold, a tolerance radius of prediction accuracy, in degrading the classification performance. This method, when combined with existing methods for sidewalk detection from aerial and street view images, could be reasonably used to help complete urban pedestrian networks in cities where high-resolution satellite images are available.

Interaction between decision-making and motor learning when selecting reach targets in the presence of bias and noise.

Motor errors can have both bias and noise components. Bias can be compensated for by adaptation and, in tasks in which the magnitude of noise varies across the environment, noise can be reduced by identifying and then acting in less noisy regions of the environment. Here we examine how these two processes interact when participants reach under a combination of an externally imposed visuomotor bias and noise. In a center-out reaching task, participants experienced noise (zero-mean random visuomotor rotations) that was target-direction dependent with a standard deviation that increased linearly from a least-noisy direction. They also experienced a constant bias, a visuomotor rotation that varied (across groups) from 0 to 40 degrees. Critically, on each trial, participants could select one of three targets to reach to, thereby allowing them to potentially select targets close to the least-noisy direction. The group who experienced no bias (0 degrees) quickly learned to select targets close to the least-noisy direction. However, groups who experienced a bias often failed to identify the least-noisy direction, even though they did partially adapt to the bias. When noise was introduced after participants experienced and adapted to a 40 degrees bias (without noise) in all directions, they exhibited an improved ability to find the least-noisy direction. We developed two models-one for reach adaptation and one for target selection-that could explain participants' adaptation and target-selection behavior. Our data and simulations indicate that there is a trade-off between adaptation and selection. Specifically, because bias learning is local, participants can improve performance, through adaptation, by always selecting targets that are closest to a chosen direction. However, this comes at the expense of improving performance, through selection, by reaching toward targets in different directions to find the least-noisy direction.

The Biological and Anthropogenic Soundscape of an Urbanized Port - the Charleston Harbor Estuary, South Carolina, USA.

Soundscape ecology provides a long-term, noninvasive approach to track animal behavior, habitat quality, and community structure over temporal and spatial scales. Using soniferous species as an indicator, biological soundscapes provide information about species and ecosystem health as well as their response and resiliency to potential stressors such as noise pollution. Charleston Harbor, South Carolina, USA provides important estuarine habitat for an abundance of marine life and is one of the busiest and fastest growing container ports in the southeast USA. Six passive acoustic recorders were deployed in the Charleston Harbor from December 2017 to June 2019 to determine biological patterns and human-associated influences on the soundscape. Anthropogenic noise was detected frequently across the estuary, especially along the shipping channel. Despite this anthropogenic noise, biological sound patterns were identified including snapping shrimp snaps (Alpheus spp. and Synalpheus spp.), fish calling and chorusing (Sciaenidae and Batrachoididae families), and bottlenose dolphin vocalizations. Biological response to anthropogenic activity varied among trophic levels, with decreased detection of fish calling when anthropogenic noise occurred and increased dolphin vocalizations in the presence of anthropogenic noise. Statistically, fine-scale, temporal patterns in biological sound were not clearly identified by sound pressure levels (SPLs), until files with anthropogenic noise presence were removed. These findings indicate that SPL patterns may be limited in their interpretation of biological activity for noisy regions and that the overall acoustic signature that we find in more pristine estuaries is lost in Charleston Harbor.

An Approach for the Reduction of Unwanted Edges in Contour Detection Based on Local Filtering

In this paper, an approach for the reduction of unwanted edges in contour detection based on local filtering is presented. Our approach can be used as a preprocessing step before contour detection. Also our approach is useful for object recognition based on feature extraction tasks, because many contour detection methods can’t delete all unwanted edges carefully. Our method consists of a computational algorithm that has 7 steps. Including smoothing, edge detection, smoothing, decreasing of pixels, thresholding, local filtering, and mask creation respectively. We use smoothing for adhering neighbor edge pixels and weakening alone edge pixels. So we can amplify the correct edge pixels and attenuate unwanted edge pixels by smoothing the edge image. In local filtering, we use a proposed casual template that determines noisy regions and correct regions and therefore can create a mask matrix that its elements related to mentioned regions. Finally we can use the "mask matrix" for improving contours by using a “And” operator and we ensure final contour that has a few context effect.

Interactive Denoising of 3D Volumes Using Wavelets

Conventional signal denoising methods either focus on statistical properties or employ a (complicated) set of parameters a user needs to tweak appropriately. More recently, AI-based denoising methods are increasingly employed due to their robustness. However, these methods typically require a large number of training datasets, which are costly to produce in a computed tomography scenario. This work introduces a denoising procedure which combines classifier-based decision making with interactive user input in the wavelet domain. These interactive user inputs allows for an easy and straight-forward specification of noisy regions, which works especially well in the wavelet basis, without the need for statistical properties or large numbers of parameters. At the same time, the obtained results show that noise in the scanned data is reduced considerably while keeping the structural similarity intact.

Parameter calibration in wake effect simulation model with stochastic gradient descent and stratified sampling

As the market share of wind energy has been rapidly growing, wake effect analysis is gaining substantial attention in the wind industry. Wake effects represent a wind shade cast by upstream turbines to the downwind direction, resulting in power deficits in downstream turbines. To quantify the aggregated influence of wake effects on the power generation of a wind farm, various simulation models have been developed, including Jensen’s wake model. These models include parameters that need to be calibrated from field data. Existing calibration methods are based on surrogate models that impute the data under the assumption that physical and/or computer trials are computationally expensive, typically at the design stage. This, however, is not the case where large volumes of data can be collected during the operational stage. Motivated by the wind energy application, we develop a new calibration approach for big data settings without the need for statistical emulators. Specifically, we cast the problem into a stochastic optimization framework and employ stochastic gradient descent to iteratively refine calibration parameters using randomly selected subsets of data. We then propose a stratified sampling scheme that enables choosing more samples from noisy and influential sampling regions and thus reducing the variance of the estimated gradient for improved convergence. Through both theoretical and numerical studies on wind farm data, we highlight the benefits of our variance-conscious calibration approach.

Recolorable Posterization of Volumetric Radiance Fields Using Visibility‐Weighted Palette Extraction

AbstractVolumetric radiance fields have recently gained significant attention as promising representations of photorealistic scene reconstruction. However, the non‐photorealistic rendering of such a representation has barely been explored. In this study, we investigate the artistic posterization of the volumetric radiance fields. We extend the recent palette‐based image‐editing framework, which naturally introduces intuitive color manipulation of the posterized results, into the radiance field. Our major challenge is applying stylization effects coherently across different views. Based on the observation that computing a palette frame‐by‐frame can produce flickering, we propose pre‐computing a single palette from the volumetric radiance field covering its entire visible color. We present a method based on volumetric visibility to sample visible colors from the radiance field while avoiding occluded and noisy regions. We demonstrate our workflow by applying it to pre‐trained volumetric radiance fields with various stylization effects. We also show that our approach can produce more coherent and robust stylization effects than baseline methods that compute a palette on each rendered view.

An Approach of Deep Learning Neural Network Model for Image Security

Abstract: Securing the Image and video data have applications in various fields like internet communication, multimedia systems, medical imaging, Telemedicine, and military communication. Watermarking protects digital intellectual property. It helps to prove the origin of an image and discourages unaccredited copies or distribution. The proposed method uses the concept of binding a secret message within another, ordinary, message. Our algorithm is used to unobtrusively hide a small message within the noisy regions of a larger image. Deep neural networks are simultaneously trained to create the hiding and revealing processes and are designed to specifically work as a pair. Different models are implemented to check the performance of the algorithm. The results are evaluated using 40,000 different types of images. With the state of art techniques like hiding the image in the least significant bits of the carrier image, our approach compresses and distributes the secret image’s representation across all of the available bits