Image Denoising Research Articles

Comparative Analysis of Deep Learning-Based and Traditional Methods for Image Denoising and Restoration

Abstract: This paper presents a comprehensive comparative analysis of deep learning-based methods and traditional techniques for image denoising and restoration. The objective is to evaluate the performance, computational efficiency, and generalizability of these methods across various types and levels of noise in images. This report presents a comprehensive comparative analysis of image denoising and restoration techniques, focusing on traditional methods and deep learning-based approaches. The study evaluates the performance, computational efficiency, and generalizability of these methods across various types and levels of noise in images.

Numerical image denoising and deblurring via an approximate weighted mean curvature flow model

A new mathematical model for image denoising and deblurring is proposed and numerically implemented. It is based on a geometric differential equation that describes motion of a level surface of its solution by the weighted mean curvature. The numerical experiments are carried out to demonstrate the computational effectiveness of the proposed technique in comparison with the weighted total variation flow and VH-regularization. Предложена и численно реализована новая математическая модель устранения шума и нечеткости изображений. Она основана на геометрическом дифференциальном уравнении, которое описывает движение поверхности уровня его решения под влиянием средней кривизны с весом. Численные эксперименты проводятся с целью демонстрации вычислительной эффективности предлагаемого метода в сравнении с методом полной вариации с весом и VH-регуляризацией.

Effect Analysis of Fourier Transform and Neural Network on Image Noise Reduction

During the acquisition, transmission, and processing of digital images, noise often appears in the pictures, seriously affecting the quality of the images and the accuracy of subsequent analysis. To solve this problem, many scholars have conducted in-depth research on image denoising, aiming to find denoising methods suitable for different scenarios. This study aims to compare the effects of Fourier transform and neural network methods in eliminating stripes in pictures. The discrete Fourier transform denoising method performs best in sharpness and has a fast processing time, but its image restoration is poor. Although the artificial neural network method has a long processing time and depends on the training set, it has good noise reduction effect. The Photoshop (beta) method is relatively simple and easy to use, with good overall effect. It is hoped that through further research, a better understanding of the role of Fourier transform and neural network methods in image noise elimination can be achieved, providing more references for the development of image processing technology. This article introduces in detail the Fourier transform method and neural network method, analyzes their advantages and disadvantages and scope of application. These results shed light on guiding further exploration of noise reduction.

GDAFormer: Gradient-guided Dual Attention Transformer for Low-Dose CT image denoising

Low-Dose Computed Tomography (LDCT) image denoising is an important problem in the research field of CT, as LDCT images exhibit significant noise and artifacts compared to normal-dose CT images. To address this issue, we propose an LDCT image denoising model named Gradient-guided Dual-Attention TransFormer (GDAFormer), which adopts an encoder–decoder structure. Specifically, in the encoding part, the Gradient Denoising Module (GDM) is utilized to obtain the clean gradient map for effectively extracting edge and structure features, and they are fused with LDCT image features by Cross-Attention Transformer (CAT). The decoding part employs a Dual Self-Attention Transformer (DSAT) to capture inter-channel correlations within the fused feature maps and pixel-level correlations within individual channels, thereby improving denoising performance. Experimental results on the MayoLDCT dataset demonstrate that our proposed GDAFormer achieves competitive performance in terms of quantitative metrics and visual perceptual quality.

Back to basics: Fast denoising iterative algorithm

We introduce Back to Basics (BTB), a fast iterative algorithm for noise reduction. Our method is computationally efficient, does not require training or ground truth data, and can be applied in the presence of independent noise, as well as correlated (coherent) noise, where the noise level is unknown. We examine three study cases: natural image denoising in the presence of additive white Gaussian noise, Poisson-distributed image denoising, and speckle suppression in optical coherence tomography (OCT). Experimental results demonstrate that the proposed approach can effectively improve image quality, in challenging noise settings. Theoretical guarantees are provided for convergence stability.

Efficient Lightweight Image Denoising with Triple Attention Transformer

Transformer has shown outstanding performance on image denoising, but the existing Transformer methods for image denoising are with large model sizes and high computational complexity, which is unfriendly to resource-constrained devices. In this paper, we propose a Lightweight Image Denoising Transformer method (LIDFormer) based on Triple Multi-Dconv Head Transposed Attention (TMDTA) to boost computational efficiency. LIDFormer first implements Discrete Wavelet Transform (DWT), which transforms the input image into a low-frequency space, greatly reducing the computational complexity of image denoising. However, the low-frequency image lacks fine-feature information, which degrades the denoising performance. To handle this problem, we introduce the Complementary Periodic Feature Reusing (CPFR) scheme for aggregating the shallow-layer features and the deep-layer features. Furthermore, TMDTA is proposed to integrate global context along three dimensions, thereby enhancing the ability of global feature representation. Note that our method can be applied as a pipeline for both convolutional neural networks and Transformers. Extensive experiments on several benchmarks demonstrate that the proposed LIDFormer achieves a better trade-off between high performance and low computational complexity on real-world image denoising tasks.

Depth Sensing Imaging System Autonomous Restored Fog

Edge-preserving smoothing is an image processing technique that smooths away textures while retaining sharp edges. Image de-noising is the technique to reduce noises from corrupted images. The aim of the image denoising is to improve the contrast of the image or perception of information in images for human viewers or to provide better output for other automated image processing techniques. Outdoor images taken in hazy climate often get degraded due to the effect of haze. There are several methods to remove haze from hazy images. Most of them over saturates the dehazed images. This degrades the quality of images. The color attenuation prior technique is one of the best algorithm to remove haze from images. Based on this technique and by making use of MATLAB software, this paper suggests a simple method to remove haze from Images. The core of DSIS lies in its fusion of depth information with traditional imaging data. By leveraging depth maps obtained from sensors such as LiDAR or structured light cameras, DSIS accurately delineates objects in the scene and their respective depths. This enables selective fog removal, prioritizing objects closer to the camera for clearer representation while preserving depth cues for improved scene understanding. Our project aims to implement and evaluate DSIS in real-world scenarios, such as outdoor surveillance and automotive vision systems. Through comprehensive testing and performance analysis, we seek to demonstrate the effectiveness of DSIS in restoring visibility under varying fog densities and lighting conditions. Additionally, we will explore optimization techniques to enhance DSIS's computational efficiency, ensuring its practical feasibility for deployment in resource-constrained environments.

Distortion Correction and Denoising of Light Sheet Fluorescence Images.

Light Sheet Fluorescence Microscopy (LSFM) has emerged as a valuable tool for neurobiologists, enabling the rapid and high-quality volumetric imaging of mice brains. However, inherent artifacts and distortions introduced during the imaging process necessitate careful enhancement of LSFM images for optimal 3D reconstructions. This work aims to correct images slice by slice before reconstructing 3D volumes. Our approach involves a three-step process: firstly, the implementation of a deblurring algorithm using the work of K. Becker; secondly, an automatic contrast enhancement; and thirdly, the development of a convolutional denoising auto-encoder featuring skip connections to effectively address noise introduced by contrast enhancement, particularly excelling in handling mixed Poisson-Gaussian noise. Additionally, we tackle the challenge of axial distortion in LSFM by introducing an approach based on an auto-encoder trained on bead calibration images. The proposed pipeline demonstrates a complete solution, presenting promising results that surpass existing methods in denoising LSFM images. These advancements hold potential to significantly improve the interpretation of biological data.

Fuzzy Inference Systems to Fine-Tune a Local Eigenvector Image Smoothing Method

Image denoising is a fundamental research topic in colour image processing, analysis, and transmission. Noise is an inevitable byproduct of image acquisition and transmission, and its nature is intimately linked to the underlying processes that produce it. Gaussian noise is a particularly prevalent type of noise that necessitates effective removal while ensuring the preservation of the original image’s quality. This paper presents a colour image denoising framework that integrates fuzzy inference systems (FISs) with eigenvector analysis. This framework employs eigenvector analysis to extract relevant information from local image neighbourhoods. This information is subsequently fed into the FIS system which dynamically adjusts the intensity of the denoising process based on local characteristics. This approach recognizes that homogeneous areas may require less aggressive smoothing than detailed image regions. Images are converted from the RGB domain to an eigenvector-based space for smoothing and then converted back to the RGB domain. The effectiveness of the proposed methods is established through the application of various image quality metrics and visual comparisons against established state-of-the-art techniques.

Product of Gaussian Mixture Diffusion Models

In this work, we tackle the problem of estimating the density fX\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$ f_X $$\\end{document} of a random variable X\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$ X $$\\end{document} by successive smoothing, such that the smoothed random variable Y\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$ Y $$\\end{document} fulfills the diffusion partial differential equation (∂t-Δ1)fY(·,t)=0\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$ (\\partial _t - \\Delta _1)f_Y(\\,\\cdot \\,, t) = 0 $$\\end{document} with initial condition fY(·,0)=fX\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$ f_Y(\\,\\cdot \\,, 0) = f_X $$\\end{document}. We propose a product-of-experts-type model utilizing Gaussian mixture experts and study configurations that admit an analytic expression for fY(·,t)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$ f_Y (\\,\\cdot \\,, t) $$\\end{document}. In particular, with a focus on image processing, we derive conditions for models acting on filter, wavelet, and shearlet responses. Our construction naturally allows the model to be trained simultaneously over the entire diffusion horizon using empirical Bayes. We show numerical results for image denoising where our models are competitive while being tractable, interpretable, and having only a small number of learnable parameters. As a by-product, our models can be used for reliable noise level estimation, allowing blind denoising of images corrupted by heteroscedastic noise.

Denoising of poisson-corrupted microscopic biopsy images using fourth-order partial differential equation with ant colony optimization

Microscopic biopsies play a critical role in the identification of lung cancer. The readable features of lung such as tissue and cell are usually affected by Poisson noise. Moreover, the existence of noise reduces the quality of the image that limits the interpretation of the disease by medical specialists. To address this, an ant colony optimization based reformed fourth order partial differential equation i.e., ACO-RFPDE is proposed. In order to develop RFPDE, two major advancements have been incorporated. In first advancement, the restoration of edges and fine details preservation has been achieved by maximum likelihood estimation. In second advancement, instead of user based manual value, a mean absolute deviation of the image has been employed for automatic calculation of the edge threshold constraint of diffusion coefficient of FPDE. Furthermore, ACO is utilized for tuning of the controlling parameters of the proposed method where peak signal to noise ratio is used as a fitness function. The performance is analyzed comparatively with existing methods as well as qualitatively and quantitatively for the LC25000 dataset. The quantitative analysis includes mean square error, correlation, normalized absolute error, no reference, universal quality index, peak signal-to-noise ratio, and structural similarity index. The following values 0.175, 0.964, 0.111, 42.45, 0.946, 55.68 and 0.985 of these metrics have been obtained by the proposed method for the images of lung squamous carcinoma cell. The results of the proposed method demonstrate the successful elimination of Poisson noise while preserving the image features like edges, boundaries and enhancing the abnormal region.

Suppressing HIFU interference in ultrasound images using 1D U-Net-based neural networks

Objective. One big challenge with high-intensity focused ultrasound (HIFU) is that the intense acoustic interference generated by HIFU irradiation overwhelms the B-mode monitoring images, compromising monitoring effectiveness. This study aims to overcome this problem using a one-dimensional (1D) deep convolutional neural network. Approach. U-Net-based networks have been proven to be effective in image reconstruction and denoising, and the two-dimensional (2D) U-Net has already been investigated for suppressing HIFU interference in ultrasound monitoring images. In this study, we propose that the one-dimensional (1D) convolution in U-Net-based networks is more suitable for removing HIFU artifacts and can better recover the contaminated B-mode images compared to 2D convolution. Ex vivo and in vivo HIFU experiments were performed on a clinically equivalent ultrasound-guided HIFU platform to collect image data, and the 1D convolution in U-Net, Attention U-Net, U-Net++, and FUS-Net was applied to verify our proposal. Main results. All 1D U-Net-based networks were more effective in suppressing HIFU interference than their 2D counterparts, with over 30% improvement in terms of structural similarity (SSIM) to the uncontaminated B-mode images. Additionally, 1D U-Nets trained using ex vivo datasets demonstrated better generalization performance in in vivo experiments. Significance. These findings indicate that the utilization of 1D convolution in U-Net-based networks offers great potential in addressing the challenges of monitoring in ultrasound-guided HIFU systems.

GreenNAS: A Green Approach to the Hyperparameters Tuning in Deep Learning

This paper discusses the challenges of the hyperparameter tuning in deep learning models and proposes a green approach to the neural architecture search process that minimizes its environmental impact. The traditional approach of neural architecture search involves sweeping the entire space of possible architectures, which is computationally expensive and time-consuming. Recently, to address this issue, performance predictors have been proposed to estimate the performance of different architectures, thereby reducing the search space and speeding up the exploration process. The proposed approach aims to develop a performance predictor by training only a small percentage of the possible hyperparameter configurations. The suggested predictor can be queried to find the best configurations without training them on the dataset. Numerical examples of image denoising and classification enable us to evaluate the performance of the proposed approach in terms of performance and time complexity.

Image denoising by linear regression on non-local means algorithm

Image denoising by linear regression on non-local means algorithm

Steering Kernel Weighted Guided Image Filtering with Gradient Constraint

In order to make the weighted guided image filter (WGIF) more fully utilize the edge direction of the image, a steering kernel weighted guided image filter (SKWGIF) is proposed by employing the steering kernel to learn the edge direction and incorporating the learning results into the WGIF. However, SKWGIF does not provide a good compromise between the two possibly contradictory objectives of edge-preserving and smoothing, where the image edges are inevitably smoothed. To overcome the drawback, a SKWGIF with gradient constraint (GC-SKWGIF) is proposed by introducing the gradient constraints into the SKWGIF. The gradient constraints allow the filter to take into account the gradient variation of the edges in the filtering process, and therefore the image edges can be better preserved. To verify the effectiveness of the proposed filter, the GC-SKWGIF is applied to edge-aware smoothing, tone mapping of high range dynamic images, image denoising and haze removal. Both theoretical analysis and experimental results show that the proposed filter can produce good resultant images.

Batch-Less Stochastic Gradient Descent for Compressive Learning of Deep Regularization for Image Denoising

Batch-Less Stochastic Gradient Descent for Compressive Learning of Deep Regularization for Image Denoising

An improved nonlinear anisotropic model with p(x)‐growth conditions applied to image restoration and enhancement

This work proposes a novel nonlinear parabolic equation with ‐growth conditions for image restoration and enhancement. Based on the generalized Lebesgue and Sobolev spaces with variable exponent, we demonstrate the well‐posedness of the proposed model. As a first result, we prove the existence of a weak solution to our model when the reaction term is bounded by a suitable function. Secondly, we use the approximations method to establish the existence of a nonnegative weak SOLA (Solution Obtained as Limit of Approximations) solution to the proposed model. We illustrate our theoretical results in the context of image denoising and enhancement. Specifically, we present various numerical implementations on a set of grayscale images. To further enrich these simulations, we assess the efficiency of the proposed model on several color images. The obtained numerical results strongly suggest that our model outperforms existing state‐of‐the‐art methods in terms of both visual and quantitative comparison, demonstrating superior efficiency and robustness in noise removal and contrast enhancement.

Noise Removal from the Image Using Convolutional Neural Networks-Based Denoising Auto Encoder

With the exponential growth in the volume of digital images captured daily, there is an escalating demand for elevating image quality to achieve both accuracy and visual appeal. Addressing this need, the development of techniques for reducing image noise while preserving crucial features, such as edges, corners, and sharp structures, has become imperative. This paper delves into the significance of image denoising and introduces a novel approach utilizing a denoising autoencoder based on convolutional neural networks (CNNs). The proposed method adopts a meticulous two-step process to effectively eliminate noise. Initially, input images are segregated into training and testing sets. Subsequently, a denoising autoencoder model is trained using the designated training data. This model is then further refined through training on a CNN, enhancing its noise reduction capabilities. The evaluation of the system's performance is conducted using testing data to gauge its effectiveness. The study employs the MATLAB programming language for implementation and evaluation. Results, measured through RMSE (Root Mean Square Error) and PSNR (Peak Signal-to-Noise Ratio) criteria on two distinct datasets—the Covid19-radiography-database and SIIM-medical-images—reveal that our proposed method outperforms existing approaches significantly. This approach is particularly promising for applications demanding enhanced image quality, such as the resolution enhancement of medical images. The study contributes to the ongoing efforts in noise reduction research, offering a robust solution for improving visual perception in diverse image processing applications.

Improving the detection of hypo-vascular liver metastases in multiphase contrast-enhanced CT with slice thickness less than 5 mm using DenseNet

IntroductionThinner slices are more susceptible in detecting small lesions but suffer from higher statistical fluctuation. This work aimed to reduce image noise in multiphase contrast-enhanced CT reconstructed with slice thickness thinner than the clinical setting (i.e., 5 mm) using convolutional neural network (CNN) for enabling better detection of hypo-vascular liver metastasis. MethodsA DenseNet model was used to generate noise map for multiphase CT reconstructed with slice thickness of 2.5 mm and 1.25 mm. Image denoising was conducted by subtracting the CNN-generated noise map from CT images with reduced photon flux due to thinner slice thickness. The performance of DenseNet was evaluated on CT scans of electron density phantoms and patients with hypovascular liver metastases less than 1.5 cm in terms of Hounsfield Unit (HU) variation, statistical fluctuation, and contrast-to-noise ratio (CNR). ResultsThe phantom study demonstrated that the CNN-based denoising method was able to reduce statistical fluctuation in CT images reconstructed with slice thickness of 2.5 mm and 1.25 mm without causing significant edge blurring or variation in HU values. With regards to patient study, it was found that the denoised 2.5-mm and 1.25-mm slices had higher CNR than the conventional 5-mm slices for hypo-vascular liver metastases in all 4 phases of multiphase CT. ConclusionOur results demonstrated that the detection of hypo-vascular liver metastases in multiphase contrast-enhanced CT with slice thickness less than 5 mm could be improved by using the CNN-based denoising method. Implications for practiceReconstruction slice thickness has a strong influence on the image quality of CT imaging. A CNN-based denoising method was used in this work to reduce the image noise in multiphase contrast-enhanced CT reconstructed with slice thickness less than 5 mm.

Indirect: invertible and discrete noisy image rescaling with enhancement from case-dependent textures

Rescaling digital images for display on various devices, while simultaneously removing noise, has increasingly become a focus of attention. However, limited research has been done on a unified framework that can efficiently perform both tasks. In response, we propose INDIRECT (INvertible and Discrete noisy Image Rescaling with Enhancement from Case-dependent Textures), a novel method designed to address image denoising and rescaling jointly. INDIRECT leverages a jointly optimized framework to produce clean and visually appealing images using a lightweight model. It employs a discrete invertible network, DDR-Net, to perform rescaling and denoising through its reversible operations, efficiently mitigating the quantization errors typically encountered during downscaling. Subsequently, the Case-dependent Texture Module (CTM) is introduced to estimate missing high-frequency information, thereby recovering a clean and high-resolution image. Experimental results demonstrate that our method achieves competitive performance across three tasks: noisy image rescaling, image rescaling, and denoising, all while maintaining a relatively small model size.