Image Processing Problems Research Articles

Breaking the Haze: Enhanced CycleGAN Models with Self-Attention for Single-Image Dehazing

A common preprocessing step in vision-based applications is image dehazing, which has drawbacks such color distortion and detail loss. Conventional techniques frequently fall short of sufficiently retrieving sharp images from foggy situations. In order to improve single-image dehazing, this paper presents a unique method that combines Self-Attention Generative Adversarial Networks (SAGANs) with Cycle-Consistent Generative Adversarial Networks (CycleGAN). In order to capture long-range dependencies and enhance the perceptual quality of the dehazed images, our method includes self-attention processes from SAGANs and makes use of CycleGAN's strengths for efficient image-to-image translation without the need for paired training samples. We present a dual-framework design in which SAGANs improve textural and contextual details by focusing on relevant characteristics throughout the whole image space, while CycleGAN assures content fidelity through cycle consistency. We show through our studies that this integrated strategy achieves higher scores in common metrics like Peak Signal-to-Noise Ratio (PSNR) of 34.146 dB and Structural Similarity Index (SSIM) of 0.964, greatly outperforming both qualitative and quantitative assessments of previous methods. Together with establishing a new standard for single-image dehazing, this work shows how effective hybrid generative models can be when handling challenging image processing problems.

Laplacian edge detection method based on a new approximation of the second-order fractional derivative in the Caputo-Fabrizio sense

Edge detection is an important issue in image processing and computer vision problems. It includes all mathematical methods that aim to identify the discontinuous points in the image. This process leads to construct curves that indicate the boundaries of an object and therefore extract feature information. In recent years, this task has received great attention from researchers and scientists, as we find in the literature many methods with their different algorithms. Up to now, there are two methods, the gradient and Laplacian methods. The first depends on the first derivative, where pixels with a high gradient are considered edges, whereas the second depends on the second derivative, such as the pixel whose second derivative is zero, it is marked as an edge point. Although these methods have achieved remarkable success, they are sensitive to noise and sometimes produce false edges. For this reason, in this paper, we propose an efficient edge detection method based on the fractional Laplacian operator in the Caputo-Fabrizio sense. In this method we develop a new approximation formula to the second-order with the Caputo-Fabrizio fractional derivative. Then, we construct a new fractional mask to compute the Laplacian convolution. To check the effectiveness of our proposed method, we have provided some test images that are affected by a Gaussian noise. The numerical results and visual observation show that our proposed edge detection method gives better performance compared to the classical Laplacian method in terms of the peak signal to noise ratio (psnr) and fine details extraction.

Deep learning informed diffusion equation model for image denoising

AbstractImage denoising is one of the fundamental problems in image processing. Convolutional neural network (CNN) based denoising approaches have achieved better performance than traditional methods, such as STROLLR and BM3D. However, CNNs can easily bring unexplainable artifacts to denoised images. In this article, a Deep Learning‐Informed Diffusion Equation (DLI‐DE) framework utilizing the image prior or the image gradient prior for image denoising is proposed. The image priors and gradient priors are learned from CNN models and used as coefficients in diffusion equations. The solution of DLI‐DE is infinitely smooth from the uniqueness of existence theorem, which guarantees that the denoised image is free of artifacts. Good properties of DLI‐DE also ensure high‐quality of denoising. The experimental analysis confirms that the denoising performance of DLI‐DE is comparable to that of contemporary CNN‐based denoising methods such as TNRD and DnCNN, while effectively preventing artifacts.

An accelerated alternating direction method of multiplier for MRI with TV regularisation

Compressed Sensing (CS) is important in the field of image processing and signal processing, and CS-Magnetic Resonance Imaging (MRI) is used to reconstruct image from undersampled k-space data. Total Variation (TV) regularisation is a common technique to improve the sparsity of image, and the Alternating Direction Multiplier Method (ADMM) plays a key role in the variational image processing problem. This paper aims to improve the quality of MRI and shorten the reconstruction time. We consider MRI to solve a linear inverse problem, we convert it into a constrained optimization problem based on TV regularisation, then an accelerated ADMM is established. Through a series of theoretical derivations, we verify that the algorithm satisfies the convergence rate of O1/k2 under the condition that one objective function is quadratically convex and the other is strongly convex. We select five undersampled templates for testing in MRI experiment and compare it with other algorithms, experimental results show that our proposed method not only improves the running speed but also gives better reconstruction results.

Convergence of Peaceman-Rachford splitting method with Bregman distance for three-block nonconvex nonseparable optimization

Abstract It is of strong theoretical significance and application prospects to explore three-block nonconvex optimization with nonseparable structure, which are often modeled for many problems in machine learning, statistics, and image and signal processing. In this article, by combining the Bregman distance and Peaceman-Rachford splitting method, we propose a novel three-block Bregman Peaceman-Rachford splitting method (3-BPRSM). Under a general assumption, global convergence is presented via optimality conditions. Furthermore, we prove strong convergence when the augmented Lagrange function satisfies Kurdyka-Łojasiewicz property. In addition, if the association function possessing the Kurdyka-Łojasiewicz property exhibits a distinctive structure, then linear and sublinear convergence rate of 3-BPRSM can be guaranteed. Finally, a preliminary numerical experiment demonstrates the effectiveness.

Comparison of different algorithms to improve the quality of Image

Super-resolution (SR) become the most important task in image processing because of requirement of high quality of images. It becomes popular in technology like computer vision and pattern recognition. In Single Image Super-resolution (SR) uses the sequences of single low resolution (LR) images from high resolution (HR) to improve the quality of image. The SR is an image processing problem which aims to generate a high resolution (HR) image from low resolution input image. The SR is difficult because of the missing information in the given LR image. Thus in SR it is very important to boost the reconstruction performance so that, there is need of effective prior knowledge. The prior knowledge of image will solve the problem to improve the quality of generated image. The main aim of SR is to generate the HR image with good visual perception as similar as original image. The paper compare two methods of SR that is bi-linear method, and direction group sparsity of image method that is proposed one. These methods will evaluate the accuracy, subjective visual effect, quality of image, time complexity and PSNR, etc. These calculations will enhance nature of picture like better visual impact quality, lower reconstruction error and adequate calculation proficiency, and so on

Image Enhancement Thanks to Negative Grey Levels in the Logarithmic Image Processing Framework.

The present study deals with image enhancement, which is a very common problem in image processing. This issue has been addressed in multiple works with different methods, most with the sole purpose of improving the perceived quality. Our goal is to propose an approach with a strong physical justification that can model the human visual system. This is why the Logarithmic Image Processing (LIP) framework was chosen. Within this model, initially dedicated to images acquired in transmission, it is possible to introduce the novel concept of negative grey levels, interpreted as light intensifiers. Such an approach permits the extension of the dynamic range of a low-light image to the full grey scale in "real-time", which means at camera speed. In addition, this method is easily generalizable to colour images and is reversible, i.e., bijective in the mathematical sense, and can be applied to images acquired in reflection thanks to the consistency of the LIP framework with human vision. Various application examples are presented, as well as prospects for extending this work.

Super-resolution radial fluctuations (SRRF) algorithm with better real-time performance

Super-resolution radial fluctuations (SRRF) is a type of statistical analysis based on fluorescence fluctuation characteristics to achieve super-resolution imaging. At present, facing the problems of slow reconstruction speed and weak real-time image processing. For dealing with these issues, parallel computing characteristics of FPGA is employed in this paper to achieve the acceleration of the algorithm. Firstly, the reconstruction process of SRRF algorithm process is split into five sub-modules, which are designed depended on the pipeline design idea. Secondly, the top-level module is designed to accomplish the sub-modules connection for executing the SRRF algorithmic process, while the reconstruction is finished by using the high-density image dataset. Finally, the single-frame image processing time is recorded and analysed in FPGA with the mainstream configuration of PC. The experimental results show that, in terms of the single frame image processing time, SRRF in FPGA is improved by 77.95% compared to PC, with real-time image processing capability under 10 ms image sampling period. In addition, the resolution-scaled Pearson coefficient (RSP) and resolution-scaled error (RSE) values of the reconstruction results in FPGA are similar to those of PC.

Parametric investigation of flow and heat transfer on the reflooding quenching of a cylinder rod in forced convection

A reflooding quenching boiling experiment in forced convection was conducted to investigate the flow and heat transfer performance of a FeCrAl cylindrical rod under various subcooling degrees and flow rates. Employing image processing and inverse heat conduction problem (IHCP) methods, the research provides insights into vapor film thickness, quenching front propagation, wall temperature cooling rate and minimum film boiling temperature. Key observations include the presence of numerous extremely small bubbles dispersing around the rod and a third quenching front appears in the middle of the rod under higher subcooling condition at 25 °C, attributed to a thinner vapor film, smaller bubbles resulting from film rupture and the inertia of fluid flow. Within the experimental range, the impact of the subcooling degree on the fluid flow and heat transfer performance during the quenching boiling process is more pronounced than that of the coolant flow rate. The evolution of the vapor film indicates a substantial reduction of 78.76 % in film thickness and a significant increase of 185.71 % in propagation velocity with an increase in subcooling degree, while an elevation in flow rate lead to a 29.84 % decrease in film thickness and a 45 % increase in propagation velocity. Heat transfer performance benefits from increased subcooling degrees throughout the entire quenching boiling process, whereas flow rate impacts the heat transfer performance of film boiling alone. Furthermore, by incorporating the impact of flow velocity, a predictive formula for Tmin is established with an error margin within ±5 %.

Generalized multilevel B-spline approximation for scattered data interpolation in image processing

This paper proposes a Generalized Multilevel B-spline Approximation (GMBA) method, which addresses scattered data interpolation problems in image processing. Mathematically, the GMBA provides a better solution for the B-spline control lattice by superimposing identical level B-splines compared with traditional Multilevel B-spline Approximation (MBA). Specifically, the GMBA allows the spacing of next control lattice to be subdivided arbitrarily or remained unchanged, which is determined by a predefined spacing set or the current error level. These improvements bring higher approximation accuracy and more flexibility for algorithm design to avoid over-fitting. In this paper, basic GMBA algorithm and its refined algorithm are compiled for image processing. Finally, six relevant cases are involved to test the GMBA, including surface approximation, image enlargement, image completion, and Salt-and-Pepper (SAP) noise removal. The experimental results show that the GMBA has better performance than the MBA in surface approximation and image processing, performs comparatively fast with the best performance on more than half of the standard test images compared with traditional algorithms, and has partially better performance even than deep learning algorithms. The GMBA can effectively recover meaningful details in images contaminated with even extremely high SAP noise level (up to 99%).

Stabilized BB projection algorithm for large-scale convex constrained nonlinear monotone equations to signal and image processing problems

The Barzilai–Borwein (BB) method is a popular and efficient gradient method for solving large-scale unconstrained optimization problems. In general, it converges much faster than the steepest descent (Cauchy) method. However, it may not converge, even when the objective function is strongly convex. To overcome this, by virtue of bounding the distance between sequential iterates, [Burdakov et al. (2019)] proposed a stabilized BB method. In this paper, by combining the stabilized BB method with a projection approach, we develop a stabilized BB projection (SBBP) method to solve nonlinear monotone equations with convex constraints. SBBP does not involve computing matrices proving the usefulness for solving large-scale problems. Its global convergence and Q-linear convergence rate are also established. We report numerical experiments on recovering sparse signals and restoring blurred images arising from compressive sensing, demonstrating the usefulness of SBBP.

On n-Dimensional Maxwell and Dirac Equations in Curved Space-Time and Its Applications in SO(P,Q) Group Theoretic Image Processing

Maxwell equations in p time and q spatial dimensions are formulated. Properties of the Green’s function for the associated (p,q)-wave operator are derived based on contour integration. The notions of electric and magnetic fields in (p,q)-dimensions is explored by analogy with four dimensional physics and the problem of far field computation of the radiation field generated by charges and currents in \((1,n-1)\) space-time is analyzed. SO(p,q) invariance properties of the Maxwell equations are deduced and used to formulate SO(p,q)-group theoretic image processing problems in (p,q)-dimensions. Dirac’s equation in (p,q)-dimensional space-time is derived using the Clifford algebra of the Dirac gamma matrices. SO(p,q)-invariance of the Dirac equation based on the spinor representation of \(SO(p,q)\) is mentioned. The Maxwell’s equations in (p,q)-dimensional curved space-time is analyzed using perturbation theory in the context of maximally symmetric spaces which play a fundamental role in \((p,q)\)-dimensional cosmological models for homogeneous and isotropic spaces. The Einstein-Maxwell equations for (p,q)-dimensional gravity and electromagnetism is studied and used to derive the equations of motion of point charges carrying mass moving under mutual gravitational and electromagnetic interactions in general relativity. Finally, Dirac’s equation in (p,q)-dimensional curved space-time interacting with the (p,q)-dimensional electromagnetic field is looked at. \(U(1)\)-Gauge, local SO(p,q) Lorentz and diffeomorphism invariance of this equation is analyzed. Local \(SO(p,q)\) invariance of the curved space-time Dirac equation is deduced based on transformation properties of the Dirac matrices under the spinor representation. The appendix presents some applications of group representation theoretic statistical image processing on a manifold to the situation when the image field is described by the six component electromagnetic field tensor on which the Lorentz group acts.

Application of generalised atomic wavelets in image processing problems

The problem of processing large data arrays is considered. Development of effective algorithms is the basis of successful application of new technologies. Complexity of the algorithm and accuracy of the results are the key quality indicators. The use of compact carrier functions leads to a decrease in processing time and complexity of the numerical algorithm. The accuracy of the data representation and the correctness of the results mainly depend on the approximation properties. It follows that a combination of the above characteristics is necessary to create efficient algorithms for processing large data sets. Aim of article – to develop and evaluate the possibilities of application of generalised wavelets based on Kravchenko-Rvachev functions for image processing with a given quality on the basis of the analysis of the quality loss control mechanism. Discrete image compression by Kravchenko-Rvachev functions possessing such fundamental qualities as good approximation properties, high order of smoothness and existence of a basis with a local carrier in spaces of atomic functions is considered. Since the generalised up-functions have the same properties, similar compression results can be obtained using the generalised discrete atomic transform, which is based on their application, which is a promising direction in digital image processing. The presented method of image processing can be used for various purposes, including reducing the size of data before their transmission through the communication channel from the sensor to the point of reception via a communication line with limited bandwidth, for storing the received images for their further use in the centres of information processing of remote sensing of the earth, as well as for data transmission to potential users.

Evaluation of the influence of distortions of the digital object model on the result of synthetic aperture radar image modelling

An essential problem in radar image processing is the synthesis of optimal and quasi-optimal algorithms for the formation of a vector of sufficient statistics and, as a consequence, for obtaining the features used to classify an object of observation. For this purpose it is required to consider the influence of distortions of digital three-dimensional models of objects on the results of modelling images obtained for synthetic aperture radar. Aim of article – to develop and evaluate the possibilities of the method of geometrical optics in the process of application of ray tracing apparatus for the cases of diffuse and specular reflection of waves from the surface of a reflecting object while evaluating the influence of angular distortions of a digital 3D model of an object on the results of modelling. The method of taking into account the theoretical relationship between the magnitude of coordinate distortions of the digital model and the magnitude of distortion of the received radar image is proposed. The paper evaluates the influence of distortions of digital models of three-dimensional (3D) objects on the results of modelling images obtained for synthetic aperture radar. The presented method of taking into account the influence of distortions of the digital 3D model of the object on the results of radar image modelling allows to estimate the conditions of applicability of the ray tracing method.

Data generation using generative adversarial networks to increase data volume

The article is an in-depth analysis of two leading approaches in the field of generative modeling: generative adversarial networks (GANs) and the pixel-to-pixel (Pix2Pix) image translation model. Given the growing interest in automation and improved image processing, the authors focus on the key operating principles of each model, analyzing their unique characteristics and features. The article also explores in detail the various applications of these approaches, highlighting their impact on modern research in computer vision and artificial intelligence. The purpose of the study is to provide readers with a scientific understanding of the effectiveness and potential of each of the models, and to highlight the opportunities and limitations of their application. The authors strive not only to cover the technical aspects of the models, but also to provide a broad overview of their impact on various industries, including medicine, the arts, and solving real-world problems in image processing. In addition, we have identified prospects for the use of these technologies in various fields, such as medicine, design, art, entertainment, and in unmanned aerial vehicle systems. The ability of GANs and Pix2Pix to adapt to a variety of tasks and produce high-quality results opens up broad prospects for industry and research.

Neural network big data fusion in remote sensing image processing technology

Abstract Remote sensing (RS) image processing has made significant progress in the past few years, but it still faces some problems such as the difficulty in processing large-scale RS image data, difficulty in recognizing complex background, and low accuracy and efficiency of processing. In order to improve the existing problems in RS image processing, this study dealt with ConvNext-convolutional neural network (CNN) and big data (BD) in parallel. Moreover, it combined the existing RS image processing with the high dimensional analysis of data and other technologies. In this process, the parallel processing of large data and high-dimensional data analysis technology improves the difficulty and low efficiency of large-scale RS image data processing in the preprocessing stage. The ConvNext-CNN optimizes the two modules of feature extraction and object detection in RS image processing, which improves the difficult problem of complex background recognition and improves the accuracy of RS image processing. At the same time, the performance of RS image processing technology after neural networks (NNs) and BD fusion and traditional RS image processing technology in many aspects are analyzed by experiments. In this study, traditional RS image processing and RS image processing combined with NN and BD were used to process 2,328 sample datasets. The image processing accuracy and recall rate of traditional RS image processing technology were 81 and 82%, respectively, and the F1 score was about 0.81 (F1 value is the reconciled average of accuracy and recall, a metric that combines accuracy and recall to evaluate the quality of the results, a higher F1 value indicates a better overall performance of the retrieval system). The accuracy rate and recall rate of RS image processing technology, which integrates NN and BD, were 97 and 98%, respectively, and its F1 score was about 0.97. After analyzing the process of these experiments and the final output results, it can be determined that the RS image processing technology combined with NN and BD can improve the problems of large-scale data processing difficulty, recognition difficulty under complex background, low processing accuracy and efficiency. In this study, the RS image processing technology combined with NN and BD has stronger adaptability with the help of NN and BD technology, and can adjust parameters and can be applied in more tasks.

The Effectiveness of two Dimensional Haar Wavelet Image De-noising performance using Soft or Hard Thresholding Approach

Image de-noising and restoration represent basic problems in image processing with many different applications including engineering, reconstruction of missing data during their transmission and enhancement ..etc. this work is aimed at developing effective algorithm for denoising image using new strategy algorithm of wavelet techniques ,by applying two dimensions wavelet transform using Haar wavelet. Wavelets are hierarchically decomposing mathematical tools. A noisy picture is sent to the Haar wavelet transform to create four decomposed bands(for each level), and the noise is then removed by applying a threshold to the resultant image band. Then, an inverse transform is used to provide a noise-reduced picture. Performance characteristics are improved by the suggested design in contrast to those of the current approaches.

Klasik ve Bulanık Kenar Algılama Yöntemlerinin Karşılaştırılması

Edge detection is one of the challenging problems in image processing. Four different classical edge detection methods—Sobel, Prewitt, Roberts, and Canny—and type-1 and type-2 fuzzy logic-based edge detection methods are applied to analyze two separate datasets with various properties. The datasets are STARE which contains medical images of the retina and BIPED which contains images of the street. Furthermore, two separate hybrid fuzzy logic methods are implemented. The type-1 and type-2 fuzzy inference techniques are combined to produce the hybrid-1 and hybrid-2 approaches, using the "AND" and "OR" logic operators. We compare the simulation results for each technique using three different image quality metrics. These are Mean Square Error (MSE), Peak Signal Noise Ratio (PSNR), and Structural Similarity Index (SSIM). The type-2 fuzzy technique outperformed the hybrid-1 fuzzy method in visual quality metrics comparison, demonstrating superior blood vessel recognition on the STARE retinal image dataset—a dataset that more closely resembles the human visual system. Using the BIPED street image dataset, the hybrid-1 fuzzy approach outperformed the Roberts method. The hybrid-1 fuzzy technique showed good results in the second order for both kinds of datasets. Any data and general applications can take advantage of it.

A Grassmann manifold handbook: basic geometry and computational aspects

The Grassmann manifold of linear subspaces is important for the mathematical modelling of a multitude of applications, ranging from problems in machine learning, computer vision and image processing to low-rank matrix optimization problems, dynamic low-rank decompositions and model reduction. With this mostly expository work, we aim to provide a collection of the essential facts and formulae on the geometry of the Grassmann manifold in a fashion that is fit for tackling the aforementioned problems with matrix-based algorithms. Moreover, we expose the Grassmann geometry both from the approach of representing subspaces with orthogonal projectors and when viewed as a quotient space of the orthogonal group, where subspaces are identified as equivalence classes of (orthogonal) bases. This bridges the associated research tracks and allows for an easy transition between these two approaches. Original contributions include a modified algorithm for computing the Riemannian logarithm map on the Grassmannian that is advantageous numerically but also allows for a more elementary, yet more complete description of the cut locus and the conjugate points. We also derive a formula for parallel transport along geodesics in the orthogonal projector perspective, formulae for the derivative of the exponential map, as well as a formula for Jacobi fields vanishing at one point.

QARV: Quantization-Aware ResNet VAE for Lossy Image Compression.

This paper addresses the problem of lossy image compression, a fundamental problem in image processing and information theory that is involved in many real-world applications. We start by reviewing the framework of variational autoencoders (VAEs), a powerful class of generative probabilistic models that has a deep connection to lossy compression. Based on VAEs, we develop a new scheme for lossy image compression, which we name quantization-aware ResNet VAE (QARV). Our method incorporates a hierarchical VAE architecture integrated with test-time quantization and quantization-aware training, without which efficient entropy coding would not be possible. In addition, we design the neural network architecture of QARV specifically for fast decoding and propose an adaptive normalization operation for variable-rate compression. Extensive experiments are conducted, and results show that QARV achieves variable-rate compression, high-speed decoding, and better rate-distortion performance than existing baseline methods.