Image Noise Removal Research Articles

Image Noise Removal Method Based on Improved Nonlocal Mean Algorithm

Aiming at the problem of unclear images acquired in interactive systems, an improved image processing algorithm for nonlocal mean denoising is proposed. This algorithm combines the adaptive median filter algorithm with the traditional nonlocal mean algorithm, first adjusts the image window adaptively, selects the corresponding pixel weight, and then denoises the image, which can have a good filtering effect on the mixed noise. The experimental results show that, compared with the traditional nonlocal mean algorithm, the algorithm proposed in this paper has better results in the visual quality and peak signal-to-noise ratio (PSNR) of complex noise images.

Is large improvement in efficiency of impulsive noise removal in color images still possible?

The substantial improvement in the efficiency of switching filters, intended for the removal of impulsive noise within color images is described. Numerous noisy pixel detection and replacement techniques are evaluated, where the filtering performance for color images and subsequent results are assessed using statistical reasoning. Denoising efficiency for the applied detection and interpolation techniques are assessed when the location of corrupted pixels are identified by noisy pixel detection algorithms and also in the scenario when they are already known. The results show that improvement in objective quality measures can be achieved by using more robust detection techniques, combined with novel methods of corrupted pixel restoration. A significant increase in the image denoising performance is achieved for both pixel detection and interpolation, surpassing current filtering methods especially via the application of a convolutional network. The interpolation techniques used in the image inpainting methods also significantly increased the efficiency of impulsive noise removal.

Multiplicative noise removal and blind inpainting of ultrasound images based on a new variational framework

Multiplicative noise removal and blind inpainting of ultrasound images based on a new variational framework

Image noise removal using optimal deep learning-based noisy pixel identification and image enhancement

ABSTRACT Image restoration is utilised to discard image noise without demolishing edge information. A new optimisation based deep model is devised for image enhancement with restoration. Here, the noisy pixel map is detected with Deep Residual Network (DRN). DRN training is carried out using the JayaBat algorithm, which was created by combining the Jaya optimisation method and the Bat algorithm (BA). The statistical model is then used to restore the noisy pixels. The neuro fuzzy model and the Image Enhancement Conditional Generative Adversarial Network (IE-CGAN) are taken into account while enhancing images. Here, the IE-CGAN training is performed with designed CAViaRJayaBat. The CAViaRJayaBat is designed newly by combining the CAViaR and JayaBat algorithms. The proposed CAViaRJayaBat-based IE-CGAN offered improved performance with the best Peak signal to noise ratio (PSNR) of 49.049 dB, the highest Second derivative measure of enhancement (SDME), 62.570 dB, and the highest structural similarity index (SSIM), 0.858.

Correction to: A Decision Based Neighbourhood Referred Asymmetrically Trimmed Modified Trimean for the Removal of High Density Salt and Pepper Noise in Images and Videos

A correction to this paper has been published: https://doi.org/10.1007/s11277-021-08608-8

A Decision Based Neighbourhood Referred Asymmetrically Trimmed Modified Trimean for the Removal of High Density Salt and Pepper Noise in Images and Videos

A Decision Based Neighbourhood Referred Asymmetrically Trimmed Modified Trimean for the Removal of High Density salt and pepper noise in Images and videos is proposed. The proposed algorithm initially checks for the outliers in a 3 × 3 neighbourhood. If the processed pixel is noisy then check for the presence of noisy pixels with the 4 neighbours; If the 4 neighbours are found to hold outliers then mean of the 4 neighbours will replace the output. If the 4 neighbours are not noisy then the output is replaced by asymmetrically trimmed Modified Trimean. If all the pixels of the current processing window are noisy then the mean of all elements will replace the processed pixel. If the processed pixel does not hold the outlier then the pixel is termed as not noisy and left unaltered. The proposed algorithm exhibit excellent noise elimination capability with enhanced edge preservation capability. The algorithm was tested on a standard database and the results of the proposed algorithm were compared to 16standard and existing algorithms. The proposed algorithm exhibit excellent results in terms of both Quantitative and qualitative measures.

An Image Enhancement Method Based on Partial Differential Equations to Improve Dark Channel Theory

Image applications appears in all aspects of life, such as medical imaging, electronic monitoring, face recognition, aerial photography remote sensing, fingerprint analysis, robot vision, automatic driving, intelligent car and so on in the civil field, as well as tanks, armored vehicles, artillery, missiles, bat submunitions in the military field. However, in the process of image acquisition or storage and transportation, the image quality will decline due to environmental conditions, equipment conditions or transmission compression, and so on. In order to accurately analyze the image, we need to carry out enhancement processing. In this paper, the current research status of image enhancement methods is analyzed. The traditional dark channel prior algorithm has some problems, such as edge or texture blur and so on. The processing effect is not ideal. Considering the limitations of the traditional dark channel prior algorithm, this paper proposes a improved algorithms based on the dark channel prior by combining partial differential equation. The main work includes the following aspects. First, the partial differential equation algorithm is combined with the dark channel prior method, and the improved algorithm is applied to enhance the image with fog. The contrast experiment of enhancement effect and running time is designed to verify the image enhancement effect of the improved algorithm. Second, the improved algorithms proposed in this paper are compared with traditional algorithms. After making the subjective and objective evaluation, it is found that the improved method in this paper has better image processing effect, stronger fog and noise removal effect, better edge retention effect, more information acquisition, and higher computational efficiency.

A Novel Pre-Processing Approach for the Denoising of Alzheimer Disease Image Dataset

Medical imaging doing an indispensable part in the area of medicine. Noise in the image is maddening as it worsens the quality of image. Thus, removal of noise is perpetually a problematic work in the images of all domain. Alzheimer’s disease, a neurological dysfunction in which destruction of cells in brain creates mental weakening and memory loss. The distinguished reason for Alzheimer’s disease is low brain activity and low blood flow. We proposed a framework for removal of noise in Alzheimer disease image using histogram equalization, thresholding, open morphological operation and a wavelet transform. This framework reduces the noise and significantly better than the existing methods used for Alzheimer disease images.

Speckle Noise Reduction for OCT Images Based on Image Style Transfer and Conditional GAN.

Raw optical coherence tomography (OCT) images typically are of low quality because speckle noise blurs retinal structures, severely compromising visual quality and degrading performances of subsequent image analysis tasks. In our previous study(Ma et al., 2018), we have developed a Conditional Generative Adversarial Network (cGAN) for speckle noise removal in OCT images collected by several commercial OCT scanners, which we collectively refer to as scanner T. In this paper, we improve the cGAN model and apply it to our in-house OCT scanner (scanner B) for speckle noise suppression. The proposed model consists of two steps: 1) We train a Cycle-Consistent GAN (CycleGAN) to learn style transfer between two OCT image datasets collected by different scanners. The purpose of the CycleGAN is to leverage the ground truth dataset created in our previous study. 2) We train a mini-cGAN model based on the PatchGAN mechanism with the ground truth dataset to suppress speckle noise in OCT images. After training, we first apply the CycleGAN model to convert raw images collected by scanner B to match the style of the images from scanner T, and subsequently use the mini-cGAN model to suppress speckle noise in the style transferred images. We evaluate the proposed method on a dataset collected by scanner B. Experimental results show that the improved model outperforms our previous method and other state-of-the-art models in speckle noise removal, retinal structure preservation and contrast enhancement.

Image denoising by nonlinear nonlocal diffusion equations

In this paper, we propose and study a nonlinear nonlocal diffusion equation of the Perona–Malik type for the removal of Gaussian noise in images. Based on the fixed-point theorem, we prove the unique solvability of the equation. We also show that solutions of the nonlocal equation converge to the solution of the local spatially regularized Perona–Malik equation if the kernel is rescaled appropriately. The new equation inherits the merit of the nonlocal method that restores details and textures of noisy images but avoids speckles and artifacts in homogeneous regions. Comparisons with other nonlocal methods for image denoising are presented.

Adaptive Graph Filtering with Intra-Patch Pixel Smoothing for Image Denoising

Graph filtering has recently attracted significant attention in signal processing. This study proposes an adaptive graph filtering method with intra-patch pixel smoothing (AWGF-PS) to improve the quality of denoised images. A detailed coefficient shrinkage algorithm is provided for the graph filter, which makes the noisy component in each graph frequency band adaptively shrink with its component significance. It replaces the traditional ideal low-pass graph filter and is better at removing image noise in the entire band. As for the design of the graph filters whose performance depends heavily on the topological structure among signals, we infer the graph from the super-pixels of similar patches by considering the smoothing attribute of intra-patch pixels. This graph fully exploits the relationship among the pixels in local areas. Moreover, it can achieve an efficient calculation of graph Fourier bases on a small-scale graph, which effectively enhances the graph filtering’s practicality. The experiments demonstrate that our AWGF-PS method suitably restores the denoised images. It outperforms several state-of-the-art model-based methods and is competitive with certain discriminative learning methods. In particular, it has more advantages in tackling images with significant noise.

An Automated Technique for Optic Disc Detection in Retinal Fundus Images Using Elephant Herding Optimization Algorithm

Background: Glaucoma and diabetic retinopathy are known to be the prime reasons for causing irrevocable blindness in the world. However, the complete vision loss can be obstructed through regular screening of the eye to detect the disorder at an early stage. Objective: In this paper, we have presented a novel nonlinear optimization technique that helps in automatic detection of the optic disc boundary from retinal fundus images thereby satisfying the anatomical constraints by using elephant herding optimization algorithm. Methods: In our approach, median filter has been used for the noise removal in retinal images. The pre-processed image is passed to the metaheuristic algorithm known as Elephant Herding Optimization algorithm. Results: The proposed technique for optic disc segmentation has been applied and tested on four standard publicly available datasets namely DRIVE, DIARETDB1, STARE and DRIONS-DB. The ground truth of the optic disc boundary has been collected from two experts of glaucoma: Expert A and Expert B who are specialists of glaucoma. The quantitative and qualitative analysis has been done to evaluate the performance of optic disc segmentation techniques. Conclusion: The proposed technique for optic disc detection helps to obtain the smooth boundaries in retinal fundus images. The aim has been successfully achieved by proposing an approach using EHO to achieve optimized solution. The effectiveness of the approach has been evaluated on for benchmark datasets and the acquired results have shown the accuracy values to be 100% for DRIVE, 100% for DIARETDB1, 99.25% for STARE and 99.99% for DRIONS-DB.

CNN and multi-feature extraction based denoising of CT images

In the past decades, Computed Tomography (CT) images have been widely used and played a critically role in medical diagnosis. In low-dose CT images, reducing the radiation dose can reduce damage to patients, but at the same time, the projected image is contaminated with noise, resulting in a lot of noise in the reconstructed medical CT image, which can affect the clinical diagnosis. Based on the network structure of GoogleNet and Inception series, and combined with deep residual learning and convolutional neural networks, a novel denoising method with multi-feature extraction is proposed in this paper for medical CT images. The extraction of shallow multi-features in medical CT images is achieved by combining convolution filters of different sizes. This is useful for obtaining more detailed feature information in the image. Through the fusion of image features, the noise learning in medical CT images is realized. The developed neural network is more targeted to the removal of noise in the medical CT image. Experimental results show the denoised medical CT images can better retain edge and texture area details with the proposed method. Compared with the existing methods, the method proposed in this paper improves the denoising effect of medical CT images.

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.

Significance driven inverse distance weighted filter to restore impulsive noise corrupted X-ray image

This paper presents a novel significance driven inverse distance weighted (SDIDW) filter for the impulsive noise removal in the X-ray images. The proposed SDIDW filter restores the noisy pixel using minimum number of nearest noise-free pixels to achieve good estimation while exhibiting low computational complexity. In the proposed filter, higher priority (weight) is given to nearest pixels compared to distant pixels and only sufficient nearest noise free pixels are determined to estimate the value of noisy pixel. A high level analysis of the computation complexity at varying noise density is done which shows that proposed SDIDW filter provides significant reduction in computation complexity over the adaptive median filters. Finally, the performance of the proposed filter is evaluated and compared over the state-of-the-art impulse noise removal techniques for varying noise density (wide range 10–90% and very high noise density range 91–99%). The experimental results on medical images demonstrate significant improvement in filtered images quality by the proposed filter over the state-of-the-art filters at each sample of noise density with small computational complexity.

Image Denoising using Modified Diffusion Functions on Nonlinear Second-Order Hyperbolic Model

Partial Differential Equations have been effective in solving numerous issues in image processing and computer vision fields. Image denoising is an essential process in image processing as the existence of noise affects an image’s quality. Various nonlinear PDE-based models for removal and edge-preserving in the images have been proposed in literatures. Due to the undesirable effect problems, which are the blocky effect and the edge detectors that falsely recognize it as edges, several improved nonlinear PDE-based denoising approaches are introduced. A nonlinear second-order hyperbolic model was introduced because of the linear hyperbolic equation’s weakness that is not completely reducing the blurring effects. Besides improving image quality using the model, an appropriate diffusion function is crucial for satisfactory denoising results. This paper aims to propose a modified diffusion function to be implemented on a nonlinear second-order hyperbolic model and evaluate the model’s performance that acts as an image denoising process. A set of images with different noise levels are processed using different modified diffusion functions and implemented to preserve the image details. Then, the proposed diffusion function’s performance is evaluated using the peak signal-to-noise ratio (PSNR) and compared with the performance of existing filters that are Gaussian filter, average filter and Wiener filter. The implementation of the experiment is completed using MATLAB. As a result, the modified functions have successfully removed the different levels of noise and improve the image quality with a higher value of PSNR as compared with Gaussian filter, average filter and Wiener filter. It can be concluded that the modified diffusion function is outperforming existing filters and significantly produced a better image noise removal method with better details preservation.

Adaptive fourth-order diffusion smoothing via bilateral kernel

This paper presents a flexible and adaptive fourth-order diffusion model for image noise removal. A flexible fourth-order diffusion equation is first presented. Then, an efficient and robust contextual discontinuity indicator based on the bilateral kernel is constructed and combined with the local discontinuity indicator (local spatial gradient) to get the desired adaptive fourth-order diffusion model. The proposed model degenerates to an anisotropic equation which performs the tangent smoothing for sharpening the inter-object boundaries and reduces to an isotropic equation which implements the fast smoothing in the homogenous regions for noise removal. Moreover, it is able to overcome the staircase artifacts arisen by most of second-order diffusion models. Experimental results are provided to validate the effectiveness of the proposed model with regard to objective evaluation metrics and visual performance which outperform those of some benchmark models.

Diffusion-Driven Image Denoising Model with Texture Preservation Capabilities

Noise removal in images denotes an interesting and a relatively challenging problem that has captured the attention of many scholars. Recent denoising methods focus on simultaneously restoring noisy images and recovering their semantic features (edges and contours). But preservation of textures, which facilitate interpretation and analysis of complex images, remains an open-ended research question. Classical methods (Total variation and Perona-Malik) and image denoising approaches based on deep neural networks tend to smudge fine details of images. Results from previous studies show that these methods, in addition, can introduce undesirable artifacts into textured images. To address the challenges, we have proposed an image denoising method based on anisotropic diffusion processes. The divergence term of our method contains a diffusion kernel that depends on the evolving image and its gradient magnitude to ensure effective preservation of edges, contours, and textures. Furthermore, a regularization term has been proposed to denoise images corrupted by multiplicative noise. Empirical results demonstrate that the proposed method generates images with higher perceptual and objective qualities.

Digital analysis of polydiacetylene quality tags for contactless monitoring of milk

The incorporation of colorimetric sensors as quality indicators in food packaging is an exciting new area of research that could improve food management. The standard approach, however, demands a reliable interface between the sensor and the food and risks food contamination which is a significant consumer concern. To overcome this challenge, herein, we develop a polydiacetylene/phospholipid agarose-based sensor that encapsulates milk in the hydrogel matrix during synthesis. The chemical recognition of free fatty acids, a product of microbial spoilage of the encapsulated milk, induces a gradual blue to red color change in the sensor. We demonstrate that the new composite material exhibits the same spoilage kinetics as regular liquid milk (digital colorimetric response 28 ± 1% and 27 ± 3% respectively), indicating the agarose does not preserve the milk ingredients nor inhibit the detection mechanism of the polydiacetylene sensors. As a result, this sensor can be attached to the external surface of food packaging to provide an indirect indication of food quality without the need for contact with the milk product. The quality tags we present can be “switched” on and off using dehydration and rehydration, removing the need for in situ manufacturing and allowing storage before use. We show that the quality tags produce a similar digital colorimetric response of 21 ± 2% to indicate milk spoilage after rehydration. The color change of the quality tags could not be analyzed using absorption spectroscopy, the standard technique for polydiacetylenes, due to the opacity that milk imparts on the sample. To solve this problem, we develop digital colorimetric analysis software using the Python programming language to describe the extent of color change in polydiacetylene materials and develop a new metric termed the Digital Colorimetric Response that describes polydiacetylene response with excellent linearity (R2 = 0.96). The software is programmed to employ statistical cleaning techniques that automatically remove image noise and outliers based on a pixel’s grayscale Z-score. This new approach to sensor design increases practicality and could be extended to the contactless quality monitoring of other foods, medicines and other products whose safety or quality is jeopardized with direct sensor contact.

Automatic Recognition and Correction of Volleyball Players’ Release Angle Based on Feature Statistics

The constant reform of the competition rules has promoted the innovation of volleyball techniques and tactics. In order to improve the training efficiency and competitive level of volleyball players, this study designed a volleyball player shooting angle automatic recognition and correction method based on the process of feature statistics. Firstly, the basic structure of the information acquisition system is analyzed, and the acquisition process is determined. Then, grayscale and binarization operations are carried out for color‐moving images to separate their foreground and background, and a median filtering algorithm is used to remove the image noise. Then, the image pyramid of different sizes is generated by the filter. Based on setting the datum direction, the feature of volleyball shooting is extracted by using the line formula. On this basis, we construct a support vector machine (SVM) classifier to statistically classify the features, use the histogram additive kernel support vector machine method to obtain the lens angle recognition results, and correct the lens angle through feature point matching. Simulation experiments show that this method can effectively remove image noise and make the image signal‐to‐noise ratio higher, and it can effectively identify whether volleyball players’ release Angle is correct, to achieve the purpose of timely correction.