Gradient Information Research Articles

Enhanced Scratch Detection for Textured Materials Based on Optimized Photometric Stereo Vision and Fast Fourier Transform–Gabor Filtering

In the process of scratch defect detection in textured materials, there are often problems of low efficiency in traditional manual detection, large errors in machine vision, and difficulty in distinguishing defective scratches from the background texture. In order to solve these problems, we developed an enhanced scratch defect detection system for textured materials based on optimized photometric stereo vision and FFT-Gabor filtering. We designed and optimized a novel hemispherical image acquisition device that allows for selective lighting angles. This device integrates images captured under multiple light sources to obtain richer surface gradient information for textured materials, overcoming issues caused by high reflections or dark shadows under a single light source angle. At the same time, for the textured material, scratches and a textured background are difficult to distinguish; therefore, we introduced a Gabor filter-based convolution kernel, leveraging the fast Fourier transform (FFT), to perform convolution operations and spatial domain phase subtraction. This process effectively enhances the defect information while suppressing the textured background. The effectiveness and superiority of the proposed method were validated through material applicability experiments and comparative method evaluations using a variety of textured material samples. The results demonstrated a stable scratch capture success rate of 100% and a recognition detection success rate of 98.43% ± 1.0%.

GAN-HA: A generative adversarial network with a novel heterogeneous dual-discriminator network and a new attention-based fusion strategy for infrared and visible image fusion

Infrared and visible image fusion (IVIF) aims to preserve thermal radiation information from infrared images while integrating texture details from visible images. Thermal radiation information is mainly expressed through image intensities, while texture details are typically expressed through image gradients. However, existing dual-discriminator generative adversarial networks (GANs) often rely on two structurally identical discriminators for learning, which do not fully account for the distinct learning needs of infrared and visible image information. To this end, this paper proposes a novel GAN with a heterogeneous dual-discriminator network and an attention-based fusion strategy (GAN-HA). Specifically, recognizing the intrinsic differences between infrared and visible images, we propose, for the first time, a novel heterogeneous dual-discriminator network to simultaneously capture thermal radiation information and texture details. The two discriminators in this network are structurally different, including a salient discriminator for infrared images and a detailed discriminator for visible images. They are able to learn rich image intensity information and image gradient information, respectively. In addition, a new attention-based fusion strategy is designed in the generator to appropriately emphasize the learned information from different source images, thereby improving the information representation ability of the fusion result. In this way, the fused images generated by GAN-HA can more effectively maintain both the salience of thermal targets and the sharpness of textures. Extensive experiments on various public datasets demonstrate the superiority of GAN-HA over other state-of-the-art (SOTA) algorithms while showcasing its higher potential for practical applications.

Collision probability sliding mode guidance for spacecraft autonomous obstacle avoidance under state uncertainty

Spacecraft must possess the capability of autonomous obstacle avoidance to ensure flight safety. Currently, the available methods for obstacle avoidance guidance are primarily deterministic and will fail with large state estimation errors, which necessitates the robustness of guidance algorithms towards uncertainties. A collision probability sliding mode guidance method is presented for the autonomous obstacle avoidance of spacecraft. Based on the multiple sliding mode surfaces guidance law, a sliding mode structure is proposed to realize obstacle avoidance in the presence of state uncertainty, taking into account the collision probability constraint. First, A multiple power reaching law is designed to enhance the convergence speed of the first sliding mode surface, so that the spacecraft system reaches the target state within a finite time. Subsequently, the collision probability gradient information function is defined according to the relative relation between the spacecraft and obstacles. This function is then combined with the improved reaching law to develop the second sliding mode surface. The sliding mode surface can quantify the collision probability in real-time, and generate corresponding obstacle avoidance guidance commands to prompt the spacecraft to move away from the obstacles according to the magnitude of collision probability, which is robust to the state uncertainty. The capability of the proposed guidance algorithm is validated by a series of simulations, including scenarios involving asteroid landing and spacecraft rendezvous, and the effectiveness and robustness in the face of uncertainties are confirmed.

Deep learning-aided active subspace exploration of free-stream effects for fan-shaped film cooling

Film cooling plays a critical role in protecting engine components from high temperatures that can influence safety and performance of gas turbines. However, the process is fraught with uncertainties due to complex inflow conditions and geometrical configurations. These uncertainties significantly impact cooling effectiveness, underscoring the importance of identifying the dominant factors in a quantified manner. Traditional methods, such as the Monte Carlo approach, encounter the “curse of dimensionality,” making them computationally intensive as the number of variable increases. This study tackles these challenges by employing a deep learning strategy with a convolutional neural network (CNN) model to predict film cooling effectiveness efficiently, reducing computational loads compared to traditional computational fluid dynamics (CFD) simulations. It enables the CNN model to act as a surrogate for the CFD simulation and provides gradient information for subsequent analyses. Additionally, this study employs active subspace (AS) analysis for dimensionality reduction, identifying dominant parameters based on the trained CNN model. This approach not only enhances the speed of simulations but also provides an effective way to analyze dominant parameters and carry out optimizations. Results demonstrate a strong correlation between the CNN predictions and detailed CFD simulations, with all the mean absolute errors below 0.08, validating the model's efficacy in capturing complex cooling dynamics. The dominant analysis based on the AS method gives the blowing ratio as the most important factors, and the subsequent parameter dimension reduction and suggested optimization region are also explored by applying the low-dimensional active variable. The tested optimal factors yield higher spatially averaged cooling effectiveness above 0.35 and show reasonable pattern changes compared to the maximum and minimum results in the sample.

A hybrid gradient climbing algorithm for a swarm robot-based gas leak detector

Methane emissions from leak sources can have a negative climate impact, in addition to contributing to the risk of explosions in urban environments. These risks can be minimized by developing systems that provide for an accurate and timely detection and localization of a gas leakage point. This research used a swarm of robots to detect and locate a leakage point. The localization algorithm derives from further optimization of the gradient climbing algorithm using fireflies acting as opportunistic agents. Firefly agents are characterized by their bioluminescent communication which guides them to dynamically adjust their positions and intensities based on the quality of the gradient information available to them. The proposed research focuses on enhancing gas leak detection through the development of a hybrid gradient climbing algorithm. This algorithm integrates gradient climbing techniques with swarm intelligence, utilizing the strengths of both approaches. This simulation resulted in the hybrid algorithm leading to a reduced convergence time and path lengths when compared to the swarm without opportunistic agents. The suggested approach can be important especially in gas distribution systems or in areas where human intervention is considered to be unsafe.

Monomial-agnostic computation of vanishing ideals

Approximate basis computation of vanishing ideals has recently been studied extensively in computational algebra and data-driven applications such as machine learning. However, symbolic computation and the dependency on term order remain essential gaps between the two fields. In this study, we present the first monomial-agnostic basis computation, which works fully numerically with proper normalization and without term order. This is realized by gradient normalization, a newly proposed data-dependent normalization that normalizes a polynomial with the magnitude of gradients at given points. Its data-dependent nature brings various advantages: i) efficient resolution of the spurious vanishing problem, the scale-variance issue of approximately vanishing polynomials, without accessing coefficients of terms, ii) scaling-consistent basis computation, ensuring that input scaling does not lead to an essential change in the output, and iii) robustness against input perturbations, where the upper bound of error is determined only by the magnitude of the perturbations. Existing studies did not achieve any of these. As further applications of gradient information, we propose a monomial-agnostic basis reduction method and a regularization method to manage positive-dimensional ideals.

Symmetrical Siamese Network for pose-guided person synthesis

Pose-Guided Person Image Synthesis (PGPIS) aims to generate a realistic person image that preserves the appearance of the source person while adopting the target pose. Various appearances and drastic pose changes make this task highly challenging. Due to the insufficient utilization of paired data, existing models face difficulties in accurately preserving the source appearance details and high-frequency textures in the generated images. Meanwhile, although current popular AdaIN-based methods are advantageous in handling drastic pose changes, they struggle to capture diverse clothing shapes imposed by the limitation of global feature statistics. To address these issues, we propose a novel Symmetrical Siamese Network (SSNet) for PGPIS, which consists of two synergistic symmetrical generative branches that leverage prior knowledge of paired data to comprehensively exploit appearance details. For feature integration, we propose a Style Matching Module (SMM) to transfer multi-level region appearance styles and gradient information to the desired pose for enriching the high-frequency textures. Furthermore, to overcome the limitation of global feature statistics, a Spatial Attention Module (SAM) is introduced to complement the SMM for capturing clothing shapes. Extensive experiments show the effectiveness of our SSNet, achieving state-of-the-art results on public datasets. Moreover, our SSNet can also edit the source appearance attributes, making it versatile in wider application scenarios.

Infrared and visible image fusion based on semi-global weighted least squares and guided edge-aware filters

Most of the existing image fusion methods are dedicated to extracting the respective private features in the source image to reconstruct to get a fused image that contains rich information. However, its neglect of the enhancement of texture details in the source image as well as the preservation of background structures, which makes the fusion result look less impressive. To this end, an infrared and visible image fusion method based on a semi-global weighted least squares (SGWLS) method and guided edge-aware filter (GEAF) is proposed. First, a SGWLS method is applied to decompose the source image into base and detail layers containing different scale information. Then, the detail layer are texture-enhanced by using the scale coefficients. Due to the lack of edge features in the original base layer, for this reason, a guided edge-aware filter is designed, in which the enhanced detail layer was used as the guidance image as a way to perform a quadratic feature enhancement fusion for the base layer in order to retain richer gradient information. Finally, reconstruction of the fused detail and base layers by using inverse transformation to obtain the fusion result. Compared with nine state-of-the-art fusion algorithms, a large number of experimental results demonstrate the superior performance of the proposed method in highlighting texture details and background features, as well as the obvious advantages in SF and AG metrics.

Optimization Techniques in the Localization Problem: A Survey on Recent Advances

Optimization is a mathematical discipline or tool suitable for minimizing or maximizing a function. It has been largely used in every scientific field to solve problems where it is necessary to find a local or global optimum. In the engineering field of localization, optimization has been adopted too, and in the literature, there are several proposals and applications that have been presented. In the first part of this article, the optimization problem is presented by considering the subject from a purely theoretical point of view and both single objective (SO) optimization and multi-objective (MO) optimization problems are defined. Additionally, it is reported how local and global optimization problems can be tackled differently, and the main characteristics of the related algorithms are outlined. In the second part of the article, extensive research about local and global localization algorithms is reported and some optimization methods for local and global optimum algorithms, such as the Gauss–Newton method, Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Differential Evolution (DE), and so on, are presented; for each of them, the main concept on which the algorithm is based, the mathematical model, and an example of the application proposed in the literature for localization purposes are reported. Among all investigated methods, the metaheuristic algorithms, which do not exploit gradient information, are the most suitable to solve localization problems due to their flexibility and capability in solving non-convex and non-linear optimization functions.

Grabbing Path Extraction of Deep-Sea Manganese Nodules Based on Improved YOLOv5

In an effort to enhance the efficiency and accuracy of deep-sea manganese nodule grasping behavior by a manipulator, a novel approach employing an improved YOLOv5 algorithm is proposed for the extraction of the shortest paths to manganese nodules targeted by the manipulator. The loss function of YOLOv5s has been improved by integrating a dual loss function that combines IoU and NWD, resulting in better accuracy for loss calculations across different target sizes. Additionally, substituting the initial C3 module in the network backbone with a C2f module is intended to improve the flow of gradient information while reducing computational demands. Once the geometric center of the manganese nodules is identified with the improved YOLOv5 algorithm, the next step involves planning the most efficient route for the manipulator to pick up the nodules using an upgraded elite strategy ant colony algorithm. Enhancements to the ACO algorithm consist of implementing an elite strategy and progressively decreasing the number of ants in each round. This method reduces both the number of iterations and the time required for each iteration, while also preventing the occurrence of local optimal solutions. The experimental findings indicate that the improved YOLOv5s detection algorithm boosts detection accuracy by 2.3%. Furthermore, when there are fewer than 30 target planning points, the improved algorithm requires, on average, 24% fewer iterations than the ACO algorithm to determine the shortest path. Additionally, the speed of calculation for each iteration is quicker while still providing the optimal solution.

A probabilistic full waveform inversion of surface waves

AbstractOver the past decades, surface wave methods have been routinely employed to retrieve the physical characteristics of the first tens of meters of the subsurface, particularly the shear wave velocity profiles. Traditional methods rely on the application of the multichannel analysis of surface waves to invert the fundamental and higher modes of Rayleigh waves. However, the limitations affecting this approach, such as the 1D model assumption and the high degree of subjectivity when extracting the dispersion curve, motivate us to apply the elastic full‐waveform inversion, which, despite its higher computational cost, enables leveraging the complete information embedded in the recorded seismograms. Standard approaches solve the full‐waveform inversion using gradient‐based algorithms minimizing an error function, commonly measuring the misfit between observed and predicted waveforms. However, these deterministic approaches lack proper uncertainty quantification and are susceptible to get trapped in some local minima of the error function. An alternative lies in a probabilistic framework, but, in this case, we need to deal with the huge computational effort characterizing the Bayesian approach when applied to non‐linear problems associated with expensive forward modelling and large model spaces. In this work, we present a gradient‐based Markov chain Monte Carlo full‐waveform inversion where we accelerate the sampling of the posterior distribution by compressing data and model spaces through the discrete cosine transform. Additionally, a proposal is defined as a local, Gaussian approximation of the target density, constructed using the local Hessian and gradient information of the log posterior. We first validate our method through a synthetic test where the velocity model features lateral and vertical velocity variations. Then we invert a real dataset from the InterPACIFIC project. The obtained results prove the efficiency of our proposed algorithm, which demonstrates to be robust against cycle‐skipping issues and able to provide reasonable uncertainty evaluations with an affordable computational cost.

ADAC: an active domain adaptive network with progressive learning strategy for cloud detection of remote sensing imagery

ABSTRACT To enhance the adaptability and application capability of the cloud detection model in different remote sensing satellite domains, unsupervised domain adaptation methods are employed to improve the model’s robustness and transferability. However, conducting fully unlabeled training in the target domain to be adapted may impose difficulty on further improving the accuracy. It is more practical to obtain a few labeled target data for domain adaptation. Additionally, the detection of complex cloud scenes remains a persistent challenge. To address these issues, an Active Domain Adaptation network for Cloud detection (ADAC) is proposed in this paper. This network employs active learning to select a small number of representative samples for annotation, facilitating the transfer of a source domain-trained model to the target domain. Moreover, a novel Generator (G) designed for cloud segmentation is developed, where G utilizes a progressive learning strategy by dividing the network into two stages to learn gradient information, guiding cloud boundary segmentation and enabling the distinction between cloud-like objects and clouds. Furthermore, G integrates Transformer and Convolutional Neural Network to address the issues of misclassifying thin clouds. The proposed method is validated on Landsat 8 and GF-2 datasets, representing the source and target domains, respectively. The experimental results demonstrate ADAC achieved a 4% improvement in the overall accuracy with only 5% labeled training samples in the target domain, achieving the effective adaptation performance in the GF-2 domain.

Gradient Information and Regularization for Gene Expression Programming to Develop Data-Driven Physics Closure Models

Gradient Information and Regularization for Gene Expression Programming to Develop Data-Driven Physics Closure Models

Optimizing Air Contactor Design for Efficient Carbon Capture in Direct Air Capture Systems: A Multi-Disciplinary Approach

As carbon dioxide (CO2) emissions continue to rise, direct air capture (DAC) has emerged as an important technology for removing CO2 from ambient air to combat climate change. The air contactor design is a critical factor impacting DAC system performance and costs. This research describes a strategy for improving the design of a single air contactor unit to increase CO2 capture flux while minimizing costs using optimization techniques grounded in experimental data .The research focuses on a 100 m2 slab air contactor modeled utilizing cooling tower and scrubber principles. The system includes the chemical, mechanical, and structural engineering disciplines. The mass transfer coefficient, air velocity, packing depth, and capital cost per frontal area are four major design characteristics that were maximized.The model integrated requirements for low air velocity, pressure drop, and energy consumption, as well as adequate gas-liquid interaction via structured packing selection and depth.The design space was explored using a multi-objective optimization strategy based on gradient-based and heuristic methods. Initial sampling via a design of experiments methodology demonstrated response surface trends in the objective function across the design space. The response surfaces mapped the optimization objective as a function of the design variables. Using gradient information, sequential quadratic programming efficiently located an optimal design vector that met all constraints on the design variables and feasibility requirements. The constraints included bounds on mass transfer coefficient, air velocity, packing depth, and capital cost per frontal area. The optimized air contactor design achieved a CO2 capture flux of 10,440 kg/m2-yr at a cost of $ 0.068 per kg CO2 captured. The developed optimization framework maximizes contactor performance while decreasing operational costs. A multi-objective genetic algorithm approach was used to identify the trade-offs between the two competing objectives of maximized CO2 capture and minimized cost by expanding the formulation. The Pareto-optimal front of the optimization results offered numerous optimized solutions with specific DAC system setups. This methodology, as well as the optimum configurations identified, serve as the foundation for constructing larger-scale DAC systems, in future. Ongoing research aims to improve solution approaches by including advanced optimization software packages such as DAKOTA, which offers global optimization algorithms, as well as to boost model accuracy via integrating additional choice variables. Future optimization and modeling studies will include contactor interactions and the modeling of a full DAC facility to boost confidence that a globally optimal design has been established.Overall, this study illustrated an efficient optimization strategy for air contactor design and contributes to the improvement of sustainable carbon capture technology, which is crucial to controlling rising global CO2 levels. Contactor configuration optimization enables superior DAC system designs at lower costs, hence promoting the urgent deployment of negative emissions solutions.

Federated dual averaging learning algorithm with delayed gradients for composite optimization

Federated learning offers a promising paradigm for training machine learning models in distributed edge networks. This paper focuses on a critical aspect of federated learning, specifically examining the influence of communication delays between the server and edge devices. We study the problem of federated composite optimization (FCO) in the scenario where communication delays are introduced due to client-sides. To tackle this issue, we propose a federated dual averaging learning algorithm that incorporates delayed gradients. We conduct a thorough convergence analysis of our algorithm taking into account the impact of the delayed gradient information. Our analysis demonstrates that under the assumption of bounded expected random delays, the algorithm exhibits the asymptotic convergence towards the optimal value. Furthermore, we present explicit convergence rates of the proposed algorithm for fixed and random delays. Our results show that the impact of delays on FCO problems with smooth loss functions gradually becomes negligible. Finally, to demonstrate the efficacy of the proposed algorithm, we conduct numerical simulations on the FEMNIST dataset utilizing ℓ1-regularized logistic regression. These experimental results confirm our findings that the algorithm converges effectively even in the presence of delays.

Boundary-Aware Gradient Operator Network for Medical Image Segmentation.

Medical image segmentation is a crucial task in computer-aided diagnosis. Although convolutional neural networks (CNNs) have made significant progress in the field of medical image segmentation, the convolution kernels of CNNs are optimized from random initialization without explicitly encoding gradient information, leading to a lack of specificity for certain features, such as blurred boundary features. Furthermore, the frequently applied down-sampling operation also loses the fine structural features in shallow layers. Therefore, we propose a boundary-aware gradient operator network (BG-Net) for medical image segmentation, in which the gradient convolution (GConv) and the boundary-aware mechanism (BAM) modules are developed to simulate image boundary features and the remote dependencies between channels. The GConv module transforms the gradient operator into a convolutional operation that can extract gradient features; it attempts to extract more features such as images boundaries and textures, thereby fully utilizing limited input to capture more features representing boundaries. In addition, the BAM can increase the amount of global contextual information while suppressing invalid information by focusing on feature dependencies and the weight ratios between channels. Thus, the boundary perception ability of BG-Net is improved. Finally, we use a multi-modal fusion mechanism to effectively fuse lightweight gradient convolution and U-shaped branch features into a multilevel feature, enabling global dependencies and low-level spatial details to be effectively captured in a shallower manner. We conduct extensive experiments on eight datasets that broadly cover medical images to evaluate the effectiveness of the proposed BG-Net. The experimental results demonstrate that BG-Net outperforms the state-of-the-art methods, particularly those focused on boundary segmentation.

Understanding Convolutional Neural Networks From Excitations.

Saliency maps have proven to be a highly efficacious approach for explicating the decisions of convolutional neural networks (CNNs). However, extant methodologies predominantly rely on gradients, which constrain their ability to explicate complex models. Furthermore, such approaches are not fully adept at leveraging negative gradient information to improve interpretive veracity. In this study, we present a novel concept, termed positive and negative excitation (PANE), which enables the direct extraction of PANE for each layer, thus enabling complete layer-by-layer information utilization sans gradients. To organize these excitations into final saliency maps, we introduce a double-chain backpropagation procedure. A comprehensive experimental evaluation, encompassing both binary classification and multiclassification tasks, was conducted to gauge the effectiveness of our proposed method. Encouragingly, the results evince that our approach offers a significant improvement over the state-of-the-art methods in terms of salient pixel removal, minor pixel removal, and inconspicuous adversarial perturbation generation guidance. In addition, we verify the correlation between PANEs.

Gradient-aware trade-off control strategy dynamic optimization for a fuel cell hybrid electric vehicle

Road gradients not only affect the actual performance of control strategies but also impact battery life due to the drastic changes in power demands. To balance battery degradation with fuel economy using gradient information, this study proposes a gradient-aware trade-off control strategy. Initially, a vehicle dynamics model and a battery degradation model are established. Based on the characteristics of known road information and remaining driving distance, state of charge planning of the battery is conducted. Subsequently, the Non-dominated Sorting Genetic Algorithm-II is applied for bi-objective optimization, yielding a set of Pareto solutions that represent different levels of energy consumption and battery degradation. Thereafter, by introducing a real-time battery degradation severity factor, an optimized bias coefficient is obtained, which adjusts in accordance with the gradient changes. Through the optimization of the bias line, the optimal bias solution set under different working conditions is determined, achieving the optimal control for power system. The fuel economy of the proposed strategy is improved by 6.8% relative to the mileage adaptive Equivalent Consumption Minimization Strategy, and the battery degradation inhibition is improved by 9.3%. After real-world conditions validation, the proposed strategy demonstrates good performance in both economic efficiency and battery life.

Very short-term wind power forecasting for real-time operation using hybrid deep learning model with optimization algorithm

This paper proposes a new hybrid deep learning model to enhance the accuracy of forecasting very short-term wind power generation. The proposed model comprises a convolutional layer, a long-short-term memory (LSTM) unit, and fully connected neural network. Convolution layer can automatically learn complicated features from the raw input, whereas the LSTM layers can retain useful information through which gradient information may flow over extended periods. To obtain the best performance from the forecasting model, a random search optimization technique has been developed for tuning hyper-parameters of the model developed. The 5 min datasets from the White Rock wind farm, Australia are used to investigate the effectiveness of the proposed model as wind farms are participating in spot electricity market. To compare the effectiveness, the proposed model is compared with the existing models, such as convolution neural network (CNN), LSTM, gated recurrent unit (GRU), bidirectional LSTM (BiLSTM), artificial neural network (ANN), and support vector machine (SVM). The root-mean-square error (RMSE), mean absolute error (MAE), and Theil’s inequality coefficient (TIC) are used to analyze and compare the performances of the predictive models. Based on RMSE and MAE, the proposed model exhibits a higher accuracy of approximately 23.79% and 28.63% compared to other forecasting methods, respectively.

Slim-YOLOv8: A fast and accurate algorithm for surface defect detection of steel strips

Steel strip is an extremely important industrial material and is widely used in various industrial fields. During the production process, surface defects need to be detected quickly and accurately. This study proposes a model. This study proposes a new slim-YOLOv8 (lightweight YOLOv8) detection model. The model is based on YOLOv8 and adopts a lightweight design paradigm, which reduces the number of parameters of the model and enhances the detection real-time performance. At the same time, an online reparameterization method is introduced to enhance the feature extraction capability of the network without raising the inference cost, and to improve the model's detection accuracy for complex defects. Finally, an auxiliary training head that can provide richer gradient information is added to the model to help train the model while preventing model overfitting. The performance of slim-YOLOv8 in mean average precision and parameters was evaluated on the well-known steel strip surface defect detection dataset NEU-DET, reaching a mAP of 85.8% at IoU 0.50 and 50.3% in the IoU 0.50–0.95 range. This is an improvement of 8.3% and 3%, respectively, compared to the baseline model. Meanwhile, the number of parameters of the model was reduced from 3.0 M to 2.7 M, which is 7% lower than the baseline model. The experimental results show that slim-YOLOv8 uses a smaller number of parameters, but has higher accuracy and is able to detect various defects in the dataset well.