检索结果-内蒙古大学图书馆

2025 IEEE International conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Qiao, Xiaoyu Li, Weisheng Xiao, Bin Huang, Yuping Yang, Lijian Chongqing Key Laboratory of Image Cognition Chongqing University of Posts and Telecommunications Chongqing China Key Laboratory of Cyberspace Big Data Intelligent Security Chongqing University of Posts and Telecommunications Ministry of Education Chongqing China

ISBN: (纸本)9798350368741

In accelerated MRI reconstruction problem, directly recovering all the missing k-space data from undersampled measurements is highly ill-posed and often leads to suboptimal performance. To address the problem, we propose a novel deep unfolding network (DUN) with subsampling decomposition (SD) based k-space refinement to mitigate the ill-posedness. Our method employs a parallel network architecture with a primary branch unfolded by gradient descent-inspired optimization process (GD-PB) for reconstruction. Additionally, we introduce an SD-based auxiliary branch (SD-AB) that decompose the inverse problem into moderately corrupted subproblems. We design a novel subsampling mask predictor that captures both global and local spatial correlations in k-space, ensuring the SD-AB effectively preserves the most well-reconstructed subsets as reliable region (RR). The RR in SD-AB is used to periodically refine the intermediate outputs of the GD-PB, achieving improved accuracy. Experimental results reveal that our method significantly outperforms conventional and SD-based DUN techniques, achieving superior PSNR and SSIM results compared with cutting-edge methods. © 2025 IEEE.

关键词： compressed sensing deep unfolded network MRI reconstruction parallel imaging subsampling decomposition

来源：评论

学校读者我要写书评

暂无评论

3D terrain estimation from a single landscape image

3D terrain estimation from a single landscape image

引用

作者： Takahashi, Haruka Kanamori, Yoshihiro Endo, Yuki Graduate School of Science and Technology University of Tsukuba Ibaraki Tsukuba Japan

This article presents the first technique to estimate a 3D terrain model from a single landscape image. Although monocular depth estimation also offers single-image 3D reconstruction, it assigns depth only to pixels visible in the input image, resulting in an incomplete 3D terrain output. Our method generates a complete 3D terrain model as a textured height map via a three-stage framework using deep neural networks. First, to exploit the performance of pixel-aligned estimation, we estimate terrain's per-pixel depth and color free from shadows or lights in the perspective view. Second, we triangulate the RGB-D data generated in the first stage and rasterize the triangular mesh from the top view to obtain an incomplete textured height map. Finally, we inpaint the depth and color in the missing regions. Because there are many possible ways to complete the missing regions, we synthesize diverse shapes and textures during inpainting using a variational autoencoder. Qualitative and quantitative experiments reveal that our method outperforms existing methods applying a direct perspective-to-top view transform as image-to-image translation. © 2022 John Wiley & Sons Ltd.

关键词： Pixels

来源：评论

学校读者我要写书评

暂无评论

Use of Anomaly Detection and Object Detection as Basic Support in the Recognition of Outlier data in images 3rd

Use of Anomaly Detection and Object Detection as Basic Suppo...

引用

3rd International conference on Advanced Research in Technologies, Information, Innovation and Sustainability, ARTIIS 2023

作者： Vásquez, Shendry Balmore Rosero Universidad Estatal Península de Santa Elena La Libertad avenida principal Km 1.5 vía Santa Elena La Libertad Ecuador

ISBN: (纸本)9783031488573

Despite significant advances in object detection in images, the ability to identify outlier objects in an image remains an unsolved problem. Its challenges arise due to the inherent variability of an outlier object, the limitations of the object detection approach, and the subjective nature of anomalies in different contexts. In addition, traditional object detection approaches focus on identifying previously known objects, which limits their effectiveness in detecting outlier objects. An comparative analysis based on anomaly detection and object detection is then proposed in this preliminary study, techniques that under training would allow learning a "normal" representation of the data, so that they can detect anomalies when reconstructing new data instances. If the reconstruction of a new data differs significantly from the "normal" data, it could indicate the presence of an outlier object. © 2024, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

MMCNet: Multi-Modal Point Cloud Completion Network Based on Self-Projected Views

MMCNet: Multi-Modal Point Cloud Completion Network Based on ...

引用

data-driven Optimization of Complex Systems (DOCS), International conference on

作者： Ziyuan Lu Qinglong Jiao Liangdong Xu School of Computer Science and Technology Hangzhou Dianzi University Hangzhou China

ISBN: (数字)9798350377842

ISBN: (纸本)9798350377859

Obtaining defect-free point cloud data is challenging due to performance constraints of acquisition devices and unavoidable occlusion, making point cloud data completion critical. In recent years, the rapid development of deep learning technology enabling increasingly abundant research on point cloud completion methods, with many showing excellent performance. In particular, multi-modal methods have attracted attention for effectively overcoming inherent data deficiencies. This study proposes a novel Multi-Modal Point Cloud Completion Network (MMCNet) based on self-projected view. The network uses multi-view fusion, dual-way fusion, and dual attention upsampling modules to effectively complete incomplete point clouds. The multi-view fusion module fuses image features from multiple perspectives, and the dual-way fusion module fuses point cloud and image features to realize the reconstruction of the rough point cloud. Then, the dual attention upsampling module refines the rough point cloud, focusing on missing areas while maintaining structural integrity, to recover high-quality complete point clouds. Experiments on multiple standard datasets validate the superiority of our proposed MMCNet over state-of-the-art completion networks.

关键词： Point cloud compression Performance evaluation Deep learning Fuses Focusing Maintenance Complex systems image reconstruction Standards Optimization

来源：评论

学校读者我要写书评

暂无评论

Ultrasonic testing signal image processing based on improved set empirical mode decomposition 7

Ultrasonic testing signal image processing based on improved...

引用

7th International conference on Information and Computer Technologies, ICICT 2024

作者： Zhao, Siqi Peng, Yuying Dalian Neusoft University Of Information Department Of Software And Big Data Technology Dalian City China

ISBN: (纸本)9798350385625

In view of the fact that the ensemble empirical modal decomposition (EEMD) method can avoid the modal aliasing phenomenon of the EMD method, it is still lacking in noise removal and retention of effective feature signals, so an ultrasonic nondestructive testing signal processing method combining EEMD and wavelet packets is proposed. In this method, the original signal is first processed by EEMD to ensure that the modal aliasing phenomenon is eliminated. However, this will cause the loss of some characteristic signals, and the lost signals will be retrieved through wavelet packet processing to improve the signal-to-noise ratio. In the process of decomposition signal reconstruction, the correlation coefficient method is used for feature selection, and the effective defect signal with high correlation coefficient is selected for reconstruction, which can not only effectively suppress modal aliasing, but also retain more feature signals. Through the experimental verification of the actual signal, the proposed improved method obtains a better signal-to-noise ratio, a lower root mean square error and a higher correlation coefficient, which is more suitable for processing ultrasonic nondestructive testing signals. © 2024 IEEE.

关键词： Defects

来源：评论

学校读者我要写书评

暂无评论

Gaussian Mixture Model-Based Temperature Modelling for data-Driven Laser Absorption Spectroscopy Tomography

Gaussian Mixture Model-Based Temperature Modelling for Data-...

引用

IEEE International Instrumentation and Measurement Technology conference (I2MTC)

作者： Yi, Ran Fu, Yalei Liu, Chang Univ Edinburgh Sch Engn Edinburgh Midlothian Scotland

ISBN: (纸本)9798350380903;9798350380910

Laser Absorption Spectroscopy Tomography (LAST) is a non-intrusive and robust technique to image two-dimensional temperature distributions for turbulence diagnosis. Given an effective dataset, the deep learning technique has the potential to offer the temperature reconstruction with good fidelity in LAST. However, most existing deep learning-based algorithms utilize elemental Gaussian distributions for training dataset construction. These simulated temperatures cannot characterize the complex turbulence, thus leading to the reconstructions with inadequate structural details. To solve this issue, we introduce a temperature modelling method using Gaussian Mixture Model (GMM), named GMM-LAST, to simulate the dynamic characteristics of temperature distributions and maintain the inherent stochastic characteristics of the turbulence. Thus, the following deep learning-based model could learn the intricacies of real-world turbulent temperature distributions from the training set. GMM-LAST is evaluated by numerical simulation through both quantitative and qualitative analysis. The result indicates that GMM-LAST can simulate the dynamic temperature with high accuracy and sensitivity.

关键词： Temperature Imaging Gaussian Mixture Model Laser Absorption Spectroscopy Tomography Deep Learning

来源：评论

学校读者我要写书评

暂无评论

Using Hyperspectral reconstruction for Multispectral images Change Detection 7

Using Hyperspectral Reconstruction for Multispectral Images ...

引用

7th International conference on image, Vision and Computing, ICIVC 2022

作者： Liu, Song Li, Haiwei Zhang, Geng Hu, Bingliang Chen, Junyu Xi'an Institute of Optics and Precision Mechanics Chinese Academy of Sciences Key Laboratory of Spectral Imaging Technology of Cas Xi'an China University of Chinese Academy of Sciences Beijing China

ISBN: (数字)9781665467346

ISBN: (纸本)9781665467346

In recent years, using multispectral remote sensing data to change detection has been a hot issue. However, the limited number of spectral segments cannot accurately classify ground objects, which will affect the performance of change detection. To overcome this problem, in this paper, we propose a multispectral image change detection method based on a hyperspectral reconstruction network. The original multispectral data is used to generate hyperspectral data with richer spectral information, and then the generated data is used for change detection. The whole network consists of a reconstruction network and a change detection network. Based on the input multispectral data, a small amount of hyperspectral data is used to assist the training. In addition, we design a weighted loss to balance the two subnetworks in the training stage. In the test stage, the network can automatically recover hyperspectral data from multispectral data for change detection. The experimental results on two data sets show that the hyperspectral reconstruction method can improve the accuracy of multispectral images. © 2022 IEEE.

关键词： image reconstruction

来源：评论

学校读者我要写书评

暂无评论

Research on Sonar image Denoising Based on Optimized Spatial Pixel Ranking Algorithm 8

Research on Sonar Image Denoising Based on Optimized Spatial...

引用

8th International conference on Signal and image Processing, ICSIP 2023

作者： Shi, Bo Cao, Tianyu Wang, Zitao Ge, Qiqi Guo, Wenbo Shanghai Jiao Tong University Department of Automation Shanghai China

ISBN: (纸本)9798350397932

Spatial separation-based methods are applied in the field of sonar image denoising due to their sufficient performance. We analyzed the types and causes of noise in sonar images. Building upon the principles of classical spatial image denoising theory, we propose an algorithm suitable for processing salt and pepper noise. It is called the optimized Spatial Pixel Ranking (OSPR) Algorithm. It decomposes and clusters the pixels of the image from the spatial, and then sorts the image according to the pixel value. The same values will be screened and fused. Finally, the sorted and filtered values are used for image reconstruction. To evaluate its denoising performance, we compare the OSPR algorithm with six classic image denoising algorithms. Through simulation tests on images with varying noise ratios, the OSPR algorithm demonstrates an average improvement range of 1.36-6.32dB in terms of peak signal-to-noise ratio (PSNR). Simultaneously, the average structural similarity index (SSIM) improvement ranges from 0.061-0.224. These results indicate that the algorithm effectively removes noise from sonar images while preserving important edge information. Furthermore, the algorithm will be more applied in target recognition and marine data map construction. © 2023 IEEE.

关键词： Pixels

来源：评论

学校读者我要写书评

暂无评论

Deep learning of curvature features for shape completion

引用

COMPUTERS & GRAPHICS-UK 2023年第1期115卷 204-215页

作者： Hernandez-Bautista, Marina Melero, Francisco Javier Univ Granada Dept Comp Sci & Artificial Intelligence Granada 18071 Spain Univ Granada Dept Software Engn Granada 18071 Spain Andalusian Res Inst Data Sci & Computat Intelligen Jaen Spain

The paper presents a novel solution to the issue of incomplete regions in 3D meshes obtained through digitization. Traditional methods for estimating the surface of missing geometry and topology often yield unrealistic outcomes for intricate surfaces. To overcome this limitation, the paper proposes a neural network-based approach that generates points in areas where geometric information is lacking. The method employs 2D inpainting techniques on color images obtained from the original mesh parameterization and curvature values. The network used in this approach can reconstruct the curvature image, which then serves as a reference for generating a polygonal surface that closely resembles the predicted one. The paper's experiments show that the proposed method effectively fills complex holes in 3D surfaces with a high degree of naturalness and detail. This paper improves the previous work in terms of a more in-depth explanation of the different stages of the approach as well as an extended results section with exhaustive experiments. & COPY;2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://***/licenses/by-nc-nd/4.0/).

关键词： Shape completion Curvature representation Parameterization Inpainting

来源：评论

学校读者我要写书评

暂无评论

Robust image Hashing Based on Histogram reconstruction 18th

Robust Image Hashing Based on Histogram Reconstruction

引用

18th International conference on Intelligent Information Hiding and Multimedia Signal Processing, IIH-MSP 2022

作者： Jia, Yao Cui, Chen Wang, Zhenbang Tian, Shigang Wang, Shen School of Data Science and Technology Heilongjiang University Heilongjiang Harbin China College of Mathematics Physics and Information Engineering Jiaxing University Zhejiang Jiaxing China State Grid Heilongjiang Power Co. Ltd. Heilongjiang Harbin China School of Computer Science and Technology Harbin Institute of Technology Heilongjiang Harbin China

ISBN: (纸本)9789819906048

In this paper, we propose a perceptual image hashing method based on histogram reconstruction. Our method consists of three steps, pre-processing, histogram reconstruction and hash vector generation based on the reconstructed histogram. In image pre-processing, the color image is converted to a gray one, then the gray image is standardized and the standardized image is processed with a low-pass filter. Then, we reconstruct the histogram of a gray image by using the relationship between the pixels in different groups. Finally, the hash vector is extracted from the reconstructed histogram. As the experimental results show, the proposed image hashing method is robust to most of the content-preserving operations, such as JPEG compression, noise addition and rotation. Moreover, the method proposed can effectively solve the problems in traditional histogram-based methods. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Low pass filters

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：