检索结果-内蒙古大学图书馆

IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING 2024年 10卷 1069-1079页

作者： Arguello, Paula Lopez, Jhon Sanchez, Karen Hinojosa, Carlos Rojas-Morales, Fernando Arguello, Henry Univ Ind Santander Dept Comp Sci Bucaramanga 680002 Colombia King Abdullah Univ Sci & Technol Thuwal 23955 Saudi Arabia

Scene captioning consists of accurately describing the visual information using text, leveraging the capabilities of computer vision and natural language processing. However, current image captioning methods are trained on high-resolution images that may contain private information about individuals within the scene, such as facial attributes or sensitive data. This raises concerns about whether machines require high-resolution images and how we can protect the private information of the users. In this work, we aim to protect privacy in the scene captioning task by addressing the issue directly from the optics before image acquisition. Specifically, motivated by the emerging trend of integrating optics design with algorithms, we introduce a learned refractive lens into the camera to ensure privacy. Our optimized lens obscures sensitive visual attributes, such as faces, ethnicity, gender, and more, in the acquired image while extracting relevant features, enabling descriptions even from highly distorted images. By optimizing the refractive lens and a deep network architecture for image captioning end-to-end, we achieve description generation directly from our distorted images. We validate our approach with extensive simulations and hardware experiments. Our results show that we achieve a better trade-off between privacy and utility when compared to conventional non-privacy-preserving methods on the COCO dataset. For instance, our approach successfully conceals private information within the scene while achieving a BLEU-4 score of 27.0 on the COCO test set.

关键词： Deep optics privacy-preserving vision computer vision natural language processing image captioning Deep optics privacy-preserving vision computer vision natural language processing image captioning

来源：评论

学校读者我要写书评

暂无评论

SEDD: Robust Blind image Watermarking With Single Encoder And Dual Decoders

引用

COMPUTER JOURNAL 2024年第6期67卷 2390-2402页

作者： Xiang, Yuyuan Wang, Hongxia Yang, Ling He, Mingze Zhang, Fei Sichuan Univ Sch Cyber Sci & Engn Chengdu 610065 Peoples R China Sichuan Univ Key Lab Data Protect & Intelligent Management Minist Educ Chengdu 610207 Peoples R China

Blind image watermarking is regarded as a vital technology to provide copyright of digital images. Due to the rapid growth of deep neural networks, deep learning-based watermarking methods have been widely studied. However, most existing methods which adopt simple embedding and extraction structures cannot fully utilize the image features. In this paper, we propose a novel Single-Encoder-Dual-Decoder (SEDD) watermarking architecture to achieve high imperceptibility and strong robustness. Precisely, the single encoder utilizes normalizing flow to realize watermark embedding, which can effectively fuse the watermark and cover image. For watermark extraction, we introduce a parallel dual-decoder to improve the imperceptibility and extracting ability. Extensive experiments demonstrate that better watermark robustness and imperceptibility are obtained by SEDD architecture. Our method achieves a bit error rate less than 0.1% under most attacks such as JPEG compression, Gaussian blur and crop. Besides, the proposed method also obtains strong robustness under combined attacks and social platform processing.

关键词： Blind watermarking Deep learning Robustness

来源：评论

学校读者我要写书评

暂无评论

Universal filter array with memristive crossbar 24

Universal filter array with memristive crossbar

引用

IEEE 24th International Conference on Nanotechnology (NANO)

作者： Remanan, Akhila Gopi, Anitha Aswani, A. R. James, Alex Digital Univ Kerala Sch Elect Syst & Automat Veiloor India

ISBN: (纸本)9798350386257;9798350386240

This study explores the integration of memristor crossbar as filter arrays for image processing.applications exploit the various filtering techniques. Memristor crossbar arrays offer a promising platform for parallel processing.and efficient implementation of filtering operations due to their dense and scalable architecture. Configuring each column in the crossbar array to act as a filter, it becomes possible to perform multiple filtering operations simultaneously on input images. This research investigates the feasibility and performance of utilizing memristor crossbar arrays as filter arrays with different filter structures and random dropouts in image processing. This analysis focusing on the potential of memristor based reconfigurable filter arrays in advancing the field of image processing.

关键词： image processing Memristor crossbar array Filters

来源：评论

学校读者我要写书评

暂无评论

Classification of the qilou (arcade building) using a robust image processing.framework based on the Faster R-CNN with ResNet50

引用

JOURNAL OF ASIAN architecture AND BUILDING ENGINEERING 2024年第2期23卷 595-612页

作者： Li, Ming Ho Yu, Yi Wei, Hongni Chan, Ting On Sun Yat Sen Univ Sch Geog & Planning Guangzhou Peoples R China East China Normal Univ Sch Urban & Reg Sci Shanghai Peoples R China Guangdong Univ Foreign Studies Sch Business Guangzhou Peoples R China Sun Yat Sen Univ Sch Geog & Planning Guangdong Prov Key Lab Urbanizat & Geosimulat Guangzhou Peoples R China Sun Yat Sen Univ Sch Geog & Planning 135 Xigangxi Rd Guangzhou Peoples R China

Qilou (arcade building) is a particular type of Chinese historical architecture combined with western and eastern building elements, which plays a significant role in the history of modern Chinese architecture. However, the recognition and classification of the qilou mainly rely on manual inspection, suppressing the cultural dissemination and protection of qilou relics. In this paper, we present a new framework that adopts multiple image processing.algorithms and a deep learning network to automate qilou classification. First, image dataset of the qilou is enhanced based on the Contrast Limited Adaptive Histogram Equalization (CLAHE) algorithm. Then, an improved Faster R-CNN with ResNet50 (Faster R-CNN-R) is deployed for qilou image recognition. A total of 760 images captured in Guangzhou were used for training, validation, and accuracy check of the proposed framework and several contrastive networks under the same conditions. Compared to other networks, the proposed framework works better than Faster R-CNN with VGG16 (Faster R-CNN-V) and FCOS. The accuracy of the proposed framework embedded with the Faster R-CNN-R, Faster R-CNN-V, and FCOS are 80.12%, 65.17%, and 66.35%, respectively. Based on digital images captured under different lighting conditions, the proposed framework can be used to classify nine different types of qilous, with high robustness.

关键词： Qilou object detection Faster R-CNN CLAHE classification

来源：评论

学校读者我要写书评

暂无评论

Determining image Deblur, Detecting Faces, and Enhancing Appearance Using image processing.Techniques

Determining Image Deblur, Detecting Faces, and Enhancing App...

引用

IEEE International Conference on Automatic Control and Intelligent Systems (I2CACIS)

作者： Sapari, Norazliani Md Yusof, Khairul Huda Lung, Nicholas Tan Kien Rasid, Madihah Md Hussin, Siti Maherah UTM Fac Elect Engn Ctr Elect Energy Syst CEES Skudai Johor Malaysia Management & Sci Univ Fac Informat Sci & Engn Selangor Malaysia

ISBN: (纸本)9798350372113;9798350372106

This paper explores the utilization of MATLAB for digital signal processing.(DSP) techniques in image processing.tasks, focusing on image deblurring, face detection, and facial feature enhancement. Blind deconvolution methods are employed to address image blurriness, while face detection is facilitated using cascaded object detectors. Enhancements to detected facial features involve histogram equalization, smoothing filters, skin tone adjustment, and contrast enhancement techniques, followed by seamless integration using resizing methods. MATLAB serves as a robust platform for implementing and analyzing DSP algorithms, providing insights into practical solutions for common challenges in digital image processing.

关键词： MATLAB digital signal processing DSP image deblurring face detection facial feature enhancement digital image processing

来源：评论

学校读者我要写书评

暂无评论

ReAdapt: A Reconfigurable Datapath for Runtime Energy-Quality Scalable Adaptive Filters

引用

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I-REGULAR PAPERS 2023年第1期70卷 327-339页

作者： Lopes Pereira, Pedro Taua Paim, Guilherme Cesar da Costa, Eduardo Antonio Melo de Almeida, Sergio Jose Bampi, Sergio Univ Fed Rio Grande do Sul UFRGS Inst Informat INF Grad Program Microelect PGMICRO BR-91501970 Porto Alegre RS Brazil Univ Catolica Pelotas UCPel Grad Program Elect Engn & Comp BR-96015560 Pelotas RS Brazil

This paper proposes ReAdapt-a reconfigurable datapath architecture for scaling the energy-quality trade-off of adaptive filtering at runtime. The ReAdapt can dynamically select four adaptive filtering algorithms for gradating complexity levels during runtime by reconfiguring the processing.flow in its datapath and by blocking the switching activity (e.g., reducing the CMOS dynamic power) of unused modules with data-gating. The ReAdapt proposal can scale the energy-quality trade-off by choosing the following four different levels of filter algorithms complexity: 1) least mean square (LMS);2) partial update normalized LMS (PU-NLMS);3) set-membership normalized LMS (SM-NLMS);4) normalized LMS (NLMS). The ReAdapt architecture reuses common modules of each adaptive filter, resulting in a compact VLSI hardware implementation. The ReAdapt architecture operation is implemented in a case-study for interference mitigation for electroencephalogram (EEG) signal processing. The hardware synthesis results show an increase of 6.80 times in throughput and at least a reduction of 2.84 times in energy per operation compared with the state-of-the-art adaptive filters. This paper also investigates the benefits of dynamically reconfiguring the four ReAdapt operating modes at runtime for different levels of signal-to-noise ratio (SNR) for the processed signals. We also demonstrate that dynamically reconfiguring the ReAdapt operating modes during runtime results in an optimal energy-quality trade-off which is advantageous over the conventional single static mode.

关键词： Runtime Computer architecture Complexity theory Very large scale integration Signal processing.algorithms Heuristic algorithms Finite impulse response filters VLSI design energy-quality scalable filtering reconfigurable architectures adaptive filters LMS PU-NLMS SM-NLMS NLMS digital signal processing

来源：评论

学校读者我要写书评

暂无评论

image Denoising: A Comparative Study of Convolutional Neural Networks 4th

Image Denoising: A Comparative Study of Convolutional Neural...

引用

4th International Conference on digital Technologies and Applications (ICDTA)

作者： Anzal, Oumaima Guessous, Najib Ouakrim, Youssef Univ Sidi Mohamed Ben Abdellah Dept Math & Comp Sci Lab M2PA Fes 30030 Morocco

ISBN: (纸本)9783031686528;9783031686535

image processing.is a vigorous area of study that utilizes various algorithms to manipulate, analyze, and enhance digital images. image denoising is one of the crucial applications of image processing. Still, the occurrence of image noise is inevitable due to various sources, including low light conditions, high ISO settings, and transmission artifacts, necessitating the availability of denoising techniques to significantly improve visual image quality. This is particularly important in fields such as computer vision, medical imaging and remote sensing. Not only does it facilitate image analysis by retaining important details, but it also optimizes the performance of compression algorithms, improves storyteller detection. In this project, we propose an in-depth study of image denoising, focusing on the use of convolutional neural networks (CNNs). The problem of Gaussian noise will be treated by applying different levels of s (low sigma = 15, medium sigma = 25, and high sigma = 50). During this project, a full comparative analysis will be made with the three mainCNNarchitectures: DnCNN, RIDNet, and IRCNN, illustrative of the quantitative and qualitative experimental results obtained by these different approaches. In fact, these approaches have shown impressive performance in image processing.tasks, including image denoising, since they used different techniques that can be adopted in CNN, such as regularization methods, batch normalization, and residual learning.

关键词： image Denoising Convolutional Neural Networks (CNNs) DnCNN RIDNET IRCNN Noise level PSNR SSIM

来源：评论

学校读者我要写书评

暂无评论

Most Resource Efficient Matrix Vector Multiplication on FPGAs

引用

IEEE ACCESS 2023年 11卷 3881-3898页

作者： Lehnert, Alexander Holzinger, Philipp Pfenning, Simon Mueller, Ralf Reichenbach, Marc Brandenburg Univ Technol Cottbus Senftenberg Chair Comp Engn D-03046 Cottbus Germany Friedrich Alexander Univ Erlangen Nurnberg Chair Comp Architecture D-91058 Erlangen Germany Friedrich Alexander Univ Erlangen Nurnberg Inst Digital Commun D-91058 Erlangen Germany

Fast and resource-efficient inference in artificial neural networks (ANNs) is of utmost importance and drives many new developments in the area of new hardware architectures, e.g., by means of systolic arrays or algorithmic optimization such as pruning. In this paper, we present a novel method for lowering the computation effort for ANN inference utilizing ideas from information theory. Weight matrices are sliced into submatrices of logarithmic aspect ratios. These slices are then factorized. This reduces the number of required computations without compromising on fully parallel processing. We create a new hardware architecture for this dedicated purpose. We also provide a tool to map these sliced and factorized matrices efficiently to reconfigurable hardware. By comparing to the state of the art FPGA implementations, we can prove our claim by lowering hardware resources measured in look-up-tables (LUTs) by a factor of three to six. Our method does not rely on any particular property of the weight matrices of the ANN. It works for the general task of multiplying an input vector with a constant matrix and is also suitable for digital signal processing.beyond ANNs.

关键词： Neural networks Computer architecture Matrix decomposition Field programmable gate arrays Sparse matrices Signal processing.algorithms Encoding Reconfigurable architectures Computational efficiency Constant matrix multiplication neural networks computer architecture reconfigurable architectures computational efficiency

来源：评论

学校读者我要写书评

暂无评论

Acceleration of the MVS workflow using graphics processors

引用

JOURNAL OF SUPERCOMPUTING 2025年第2期81卷 1-21页

作者： Diaz-Cano, Roberto Folch, Francesc Quintana-Orti, Enrique S. Alonso-Jorda, Pedro Multiscan Technol Alicante Spain Univ Politecn Valencia Valencia Spain

Computer Vision (CV) leverages artificial intelligence to analyse digital images, offering insights for a wide range of different applications. While CV software often relies on open-source libraries such as OpenCV, it is probably more common for this software to use custom codes. Creating particular solutions stems from the very nature of the specific CV problems being addressed but, despite these particularities, there are common links at the core that are either not addressed by generic CV libraries or require significant customisation for specific applications. Understanding the nature of real problems faced by a digital image analysis use case can contribute as much as solving a generic CV problem, and this is the aim of this paper. This article addresses the problem of migrating to CUDA a part of Multiscan Vision System, a complex CV workflow utilised in a real-world, industrial use case. The primary challenge lies in minimising the overhead due to data transfers between the host and GPU (graphics processing.unit), or even within the device's memory itself. While the speed-up achieved may not rival that of other applications more suitable to GPU architecture (in particular, massively data parallel applications), the algorithms and data distribution proposed in this study effectively offload a substantial portion of the workflow to the GPU in the context of low (integer) arithmetic intensity and real-time constraints. This frees the CPU to handle other workflow components and increases the capability to incorporate more cameras, significantly boosting productivity and economic performance.

关键词： Computer vision image processing.workflow Graphics processing.units CUDA Data parallelism High performance

来源：评论

学校读者我要写书评

暂无评论

Estimation analysis of Edge and Line Detection Methods in digital image processing

Estimation analysis of Edge and Line Detection Methods in Di...

引用

Disruptive Technologies (ICDT), International Conference on

作者： Prag Singhal Shalini Gupta Pallavi Tiwari Kamal Sharma Department of Applied Science & Humanities ABES Engineering College Ghaziabad UP India Department of Applied Science & Humanities Raj Kumar Goel Institute of Technology Ghaziabad UP India Computer Science & Engineering Graphic Era Hill University Dehradun Uttarakhand India Department of Mechanical Engineering GLA University Mathura UP India

This paper provides an estimation evaluation of the edge and line detection methods in virtual photo processing. This analysis evaluates the performance of numerous facet and line detection algorithms in phrases of facet/line first-rate, accuracy, computational complexity, and robustness to noise in the context of artificial and actual-international snapshots. Mainly, it discusses the overall performance of 4 edge detection methods: Roberts, Prewitt, Sobel, and Canny; and 3 line detection techniques: Hough remodel, Linear Estimation, and Probabilistic Hough remodel. as compared to the classical algorithms, the Probabilistic Hough rework algorithm is determined to have the excellent accuracy and robustness to noise. Furthermore, a contrast of computational complexities shows that the Hough transform has the bottom complexity and computational time, even as the Linear Estimation algorithm has the best complexity and computational time. The primary outcome of this study is that the brink/line first-rate, accuracy, computational complexity, and robustness are very dependent on the sort of entered snapshots and the respective selected parameters.

关键词：

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：