检索结果-内蒙古大学图书馆

Single image denoising based on adaptive fusion dual-domain network

IET image processing 2024年第3期18卷 561-571页

作者： Jiang, Zhiwei Xue, Zhizhong Wang, Jue Hu, Yibiao Zheng, Qiufu Xidian Univ Sch Phys Xian Shaanxi Peoples R China Zhejiang Univ Technol Coll Informat Engn Hangzhou Zhejiang Peoples R China Xidian Univ Schoole Phys Xian 710000 Shaanxi Peoples R China

Deep learning-based-image denoising methods have recently achieved excellent performance by learning non-linear mapping in the spatial domain. However, these methods fail to address the noise without specific distribution because they only use features of the spatial domain. Meanwhile, existing methods that utilize features in the frequency domain fail to combine the detailed information of both domains properly for effective reconstruction and demonstrate poor generalization. Therefore, a novel adaptive fusion dual-domain network (AFDN) is introduced for single image restoration. Different from deep learning-based methods, which operate on the spatial or dual-domains in a certain order, the proposed AFDN combine the spatial-domain image and corresponding frequency-domain image as the input and use the interlacing dual-domain module with flexible adaptability to learn the relationship between spatial and frequency domains. In experimental results, the AFDN is compared with several state-of-the-art restoration methods. Quantitative results showed that the AFDN achieves enhanced effects and high index values. The code of this paper will be released at A novel adaptive fusion dual-domain network for single image restoration. Different from deep learning-based methods, which operate on the spatial or dual-domains in a certain order, the proposed adaptive fusion dual-domain network combine the spatial-domain image and corresponding frequency-domain image as the input and use the interlacing dual-domain module with flexible adaptability to learn the relationship between spatial and frequency ***

关键词： adaptive signal processing convolutional neural nets Gaussian noise image denoising

来源：评论

学校读者我要写书评

暂无评论

KAN See in the Dark

引用

IEEE signal processing LETTERS 2025年 32卷 891-895页

作者： Ning, Aoxiang Xue, Minglong He, Jinhong Song, Chengyun Chongqing Univ Technol Chongqing 400054 Peoples R China

Low-lightimage enhancement methods are difficult to fit the complex nonlinear relationship between normal and low-light images due to uneven illumination and noise effects. The recently proposed Kolmogorov-Arnold networks (KANs) feature spline-based convolutional layers and learnable activation functions, which can effectively capture nonlinear dependencies. In this paper, we design a KAN-Block based on KANs and innovatively apply it to low-light image enhancement. This method effectively alleviates the limitations of current methods constrained by linear network structures and lack of interpretability, further demonstrating the potential of KANs in low-level vision tasks. Given the poor perception of current low-light image enhancement methods and the stochastic nature of the inverse diffusion process, we further introduce frequency-domain perception for visually oriented enhancement. Extensive experiments demonstrate the competitive performance of our method on benchmark datasets.

关键词： Diffusion model frequency-domain perception Kolmogorov-Arnold networks (KANs) Kolmogorov-Arnold networks (KANs) low-light image enhance- ment

来源：评论

学校读者我要写书评

暂无评论

SAR-CDL: SAR image interpretable despeckling through convolutional dictionary learning network

引用

signal processing 2025年 233卷

作者： Zhao, Xueqing Ren, Fuquan Sun, Haibo Zhang, Yan Ma, Yue Qi, Qinghong Yanshan Univ Sch Sci Qinhuangdao 066004 Peoples R China Tangshan Univ Tangshan 063002 Peoples R China Tianjin Univ Sch Mech Engn Tianjin 300072 Peoples R China

Deep learning-based approaches have shown advantages in the task of despeckling for SAR images. However, it is still difficult to explain due to the black-box nature of deep learning. Deep unfolding methods provide an interpretable alternative to building deep neural networks, which combines traditional iterative optimization methods with deep neural networks for image recovery tasks. In this paper, we propose an unfolded deep convolutional dictionary learning framework (SAR-CDL) for SAR image despeckling. A new variational model based on convolutional dictionary for removing multiplicative noise is proposed. The alternate direction multiplier method combining deep learning method are used to optimize the variational model, which can parameterize the model by deep learning in an end-to-end learning manner and avoid the large workload of the tuning process. The performance of the proposed SAR-CDL is validated on both simulated and real SAR datasets. The experimental results show that the proposed model outperforms many state-of-the-art methods in terms of quantitative metrics and visual quality, with a stronger ability to recover the fine structure and texture of the SAR images. In addition, the proposed SAR-CDL is robust to the size of the training set and can achieve appropriate results while reducing the training dataset.

关键词： SAR images Despeckling Interpretable Convolutional dictionary learning Deep learning

来源：评论

学校读者我要写书评

暂无评论

Data-driven gradient priors integrated into blind image deblurring

引用

signal processing-image COMMUNICATION 2025年 135卷

作者： Qi, Qing Guo, Jichang Li, Chongyi Qinghai Minzu Univ Dept Phys & Elect Informat Engn Xining 810007 Qinghai Peoples R China Tianjin Univ Dept Elect & Informat Engn Tianjin Peoples R China Nanyang Technol Univ Dept Comp Sci & Engn Singapore Singapore

Blind image deblurring is a severely ill-posed task. Most existing methods focus on deep learning to learn massive data features while ignoring the vital significance of classic image structure priors. We make extensive use of the image gradient information in a data-driven way. In this paper, we present a Generative Adversarial Network (GAN) architecture based on image structure priors for blind non-uniform image deblurring. Previous image deblurring methods employ Convolutional neural Networks (CNNs) and non-blind deconvolution algorithms to predict kernel estimations and obtain deblurred images, respectively. We permeate the structure prior of images throughout the design of network architectures and target loss functions. To facilitate network optimization, we propose multi-term target loss functions aimed to supervise the generator to have images with significant structure attributes. In addition, we design a dual-discriminant mechanism for discriminating whether the image edge is clear or not. Not only image content but also the sharpness of image structures need to be discriminated. To learn image gradient features, we develop a dual-flow network that considers both the image and gradient domains to learn image gradient features. Our model directly avoids the accumulated errors caused by two steps of "kernel estimation-non-blind deconvolution". Extensive experiments on both synthetic datasets and real-world images demonstrate that our model outperforms state-of-the-art methods.

关键词： Blind image deblurring Generative Adversarial Networks Gradient priors image structures

来源：评论

学校读者我要写书评

暂无评论

CNN-based gesture recognition using raw numerical gray-scale images of surface electromyography

引用

BIOMEDICAL signal processing AND CONTROL 2025年 101卷

作者： Chen, Qingzheng Tao, Qing Zhao, Muchao Ma, Luyao Xinjiang Univ Coll Intelligent Mfg Modern Ind Sch Mech Engn Urumqi Peoples R China

The gesture recognition based on surface electromyography signals (sEMG) is an important human-computer interaction technology. This paper proposes a precise gesture recognition method that processes sEMG into raw numerical grayscale images and combines them with Convolutional neural Networks (CNN) as classifiers to address challenges such as the difficulty of manual feature extraction from sEMG and low classifier recognition accuracy. Firstly, sEMG data from four channels are collected for 13 specific hand gestures from 13 subjects. Applying linear and exponential operations to sEMG voltage values in the time domain, it maps the resulting matrix to a numerical range of 0-255 and converts it into a single-layer grayscale image. Subsequently, a Convolutional neural Network classification model (CNN4-M) is constructed, and the model is trained using these single-layer grayscale images. To assess the grayscale image method, DB1 and DB3 datasets from Ninapro were used for validation. Six classical CNNs are trained using these single-layer grayscale images, and their training results are compared to sEMG spectrograms generated using Short-Time Fourier Transform (STFT). Additionally, the gesture recognition method proposed in this paper is compared to five common machine learning methods. Results showed that the original grayscale images worked well with CNNs, outperforming STFT-generated images during validation. The gesture recognition method proposed in this paper outperforms five common machine learning methods in terms of recognition accuracy. Finally, training CNN4-M with the original grayscale images reached the highest validation accuracy, with 98.03% classification accuracy for 13 gestures and 99.95% and 98.07% on DB1 and DB3 datasets, respectively.

关键词： Gesture recognition Convolutional neural network Surface electromyography Grayscale image

来源：评论

学校读者我要写书评

暂无评论

An enhanced network with parallel graph node diffusion and node similarity contrastive loss for hyperspectral image classification

引用

DIGITAL signal processing 2025年 158卷

作者： Ye, Hailiang Huang, Xiaomei Zhu, Houying Cao, Feilong China Jiliang Univ Coll Sci Dept Appl Math Hangzhou 310018 Peoples R China Macquarie Univ Sch Math & Phys Sci Sydney Australia Zhejiang Normal Univ Sch Math Sci Jinhua 321004 Peoples R China

Graph neural networks (GNNs) have substantially advanced hyperspectral image (HSI) classification. However, GNN-based methods encounter challenges in identifying significant discriminative features with high similarity across long distances and transmitting high-order neighborhood information. Consequently, this paper proposes an enhanced network based on parallel graph node diffusion (PGNDE) for HSI classification. Its core develops a parallel multi-scale graph attention diffusion module and a node similarity contrastive loss. Specifically, the former first constructs a multi-head attention-forward propagation (AFP) module for different scales, which incorporates multi-hop contextual information into attention calculation and diffuses information in parallel throughout the network to capture critical feature information within the HSI. Afterward, it builds an adaptive weight computation layer that collaborates with multiple parallel AFP modules, enabling the adaptive calculation of node feature weights from various AFP modules and generating desired node representations. Moreover, a node similarity contrastive loss is devised to facilitate the similarity between superpixels from the same category. Experiments with several benchmark HSI datasets validate the effectiveness of PGNDAF across existing methods.

关键词： Hyperspectral image classification Graph neural networks Similarity contrastive loss Feature representation

来源：评论

学校读者我要写书评

暂无评论

RIHINNet: A robust image hiding method against JPEG compression based on invertible neural network

引用

IET image processing 2024年第13期18卷 3801-3814页

作者： Jin, Xin Pan, Chengyi Cheng, Zien Dong, Yunyun Jiang, Qian Yunnan Univ Engn Res Ctr Cyberspace Kunming 650000 Peoples R China Yunnan Univ Sch Software Kunming Peoples R China

image hiding is a task that embeds secret images in digital images without being detected. The performance of image hiding has been greatly improved by using the invertible neural network. However, current image hiding methods are less robust in the face of Joint Photographic Experts Group (JPEG) compression. The secret image cannot be extracted from the stego image after JPEG compression of the stego image. Some methods show good robustness for some certain JPEG compression quality factors but poor robustness for other common JPEG compression quality factors. An image-hiding network (RIHINNet) that is robust to all common JPEG compression quality factors is proposed. First of all, the loss function is redesigned;thus, the secret image is hidden as much as possible in the area that is less likely to be changed after JPEG compression. Second, the classifier is designed, which can help the model to select the extractor according to the range of JPEG compression degree. Finally, the interval robustness of the secret image extraction is improved through the design of a denoising module. Experimental results show that this RIHINNet outperforms other state-of-the-art image-hiding methods in the face of JPEG compressed noise with random compression quality factors, with more than 10 dB peak signal-to-noise ratio improvement in secret image recovery on imageNet, COCO and DIV2K datasets. The loss function is designed to hide the secret image as much as possible in the area that will not change much after JPEG compression, which is helpful to improve the robustness of the model against JPEG compression. The current image-hiding model fails to converge when it is trained robustly for all compression quality factors. Through the design of the classifier, the model can select the extractor according to the range of JPEG compression degree, which can make the model converge while improving the quality of the recovered secret image. The denoising module is designed to improve the

关键词： image processing neural nets steganography

来源：评论

学校读者我要写书评

暂无评论

TDU-DLNet: A transformer-based deep unfolding network for dictionary learning

引用

signal processing 2025年 231卷

作者： Wu, Kai Dong, Jing Hu, Guifu Liu, Chang Wang, Wenwu Nanjing Tech Univ Coll Elect Engn & Control Sci Nanjing Jiangsu Peoples R China Shandong Univ Sci & Technol Coll Ocean Sci & Engn Qingdao Shandong Peoples R China Univ Surrey Ctr Vis Speech & Signal Proc Guildford GU2 7XH England

Deep unfolding attempts to leverage the interpretability of traditional model-based algorithms and the learning ability of deep neural networks by unrolling model-based algorithms as neural networks. Following the framework of deep unfolding, some conventional dictionary learning algorithms have been expanded as networks. However, existing deep unfolding networks for dictionary learning are developed based on formulations with pre-defined priors, e.g., l1-norm, or learn priors using convolutional neural networks with limited receptive fields. To address these issues, we propose a transformer-based deep unfolding network for dictionary learning (TDU-DLNet). The network is developed by unrolling a general formulation of dictionary learning with an implicit prior of representation coefficients. The prior is learned by a transformer-based network where an inter-stage feature fusion module is introduced to decrease information loss among stages. The effectiveness and superiority of the proposed method are validated on image denoising. Experiments based on widely used datasets demonstrate that the proposed method achieves competitive results with fewer parameters as compared with deep learning and other deep unfolding methods.

关键词： Deep unfolding network Dictionary learning Transformer networks image denoising

来源：评论

学校读者我要写书评

暂无评论

A Review of image Enhancement methods Based on stochastic Resonance of Neuronal Networks 5

A Review of Image Enhancement Methods Based on Stochastic Re...

引用

5th IEEE International Conference on Pattern Recognition and Machine Learning, PRML 2024

作者： Sun, Ximin Niu, Pingjuan Li, Yong Li, Xiaoming Zhao, Bingzhen Zhang, Bin Liu, Dan Sun, Bo State Gride Ecommerce Technology CO. LTD Tiangong University Laboratory of Metaverse in Electric Power Industry Tian'jin China Tiangong University Laboratory of Metaverse in Electric Power Industry Tian'jin China State Gride Ecommerce Technology CO. LTD Laboratory of Metaverse in Electric Power Industry Tian'jin China

ISBN: (纸本)9798350355925

stochastic resonance has always been a research hotspot in the field of image processing. The ability of the nervous system to detect signals is closely related to nonlinear and collective behavior in neural mediators. Neuronal noise is the most significant factor influencing this ability, and it is a natural endogenous fluctuation in brain activity. Experimental and theoretical studies have shown that noise is beneficial, not harmful, for the normal functioning of the nervous system. In this regard, it is generally agreed that noise of appropriate intensity can participate in the rhythmic activity of the brain, and that noise-induced oscillations enhance the performance of weak signal processing, especially when the frequency of the signal is approximately noise-induced rhythmic oscillations. This behavior in the biological nervous system can be explained by the concept of 'stochastic resonance'. Correspondingly, the stochastic resonance mechanism of neurons is used to denoise the image, which has a good effect. In this paper, we first introduce an image processing method using random resonance of neurons. Then, the commonly used evaluation indicators in the field of image processing are sorted out. Finally, the development trend of image processing using random resonance of neurons is prospected. © 2024 IEEE.

关键词： Circuit resonance

来源：评论

学校读者我要写书评

暂无评论

Adaptive Fault-Tolerant Control for Pure-Feedback stochastic Nonlinear Systems with Sensor and Actuator Faults

引用

CIRCUITS SYSTEMS AND signal processing 2023年第10期42卷 5838-5867页

作者： Bali, Arun Chouhan, Siddharth Singh Kumar, Gourav Kumar, Rahul Singh, Uday Pratap Shri Mata Vaishno Devi Univ Sch Math Katra 182320 Jammu & Kashmir India VIT Bhopal Univ Sch Comp Sci & Engn Sehore 466114 Madhya Pradesh India Cent Univ Jammu Satish Dhawan Ctr Space Sci Samba 181143 Jammu & Kashmir India

In this article, for a class of stochastic pure-feedback nonlinear systems with simultaneous actuator and sensor faults, the problem of adaptive fault-tolerant control is examined. The stochastic pure-feedback nonlinear system is first converted into a strict-feedback by applying the mean value theorem and radial basis function neural networks are used to approximate the unknown functions. Only one adaptive parameter needs to be calculated online rather than the actual weight vector elements by determining the greatest value of the norm of the neural network weight vector. With the help of regrouping and parameter separation methods, the unavailability of state variables caused by sensor faults is addressed. The Lyapunov function methods and the backstepping recursive design technique are used to design an adaptive fault-tolerant controller. It is shown that by choosing proper the design parameters, the tracking errors converge to a small region of the origin, and all the signals in the closed-loop system are bounded in probability. The performance of the proposed controller is illustrated using a numerical example and a real-world example of a rigid robot manipulator system.

关键词： Pure-feedback nonlinear systems stochastic systems Fault-tolerant control Sensor fault Actuator fault Lyapunov function

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：