检索结果-内蒙古大学图书馆

Symposium of image, Signal processing, and Artificial Vision (STSIVA)

作者： Norikazu Takahashi Mutsuki Oura Tsuyoshi Migita Graduate School of Natural Science and Technology Okayama University Okayama Japan

ISBN: (数字)9781728170862

ISBN: (纸本)9781728170879

Principal component analysis is one of the most important methods of multivariate analysis, and has been applied in a wide range of fields such as statistical analysis, machine learning, pattern recognition, signal processing, and communication. Recently, using the idea of multi-agent networks, distributed algorithms for principal component analysis have been proposed for the case where the data matrix is partitioned either row-wise or column-wise. In this paper, considering the case where the data matrix is partitioned both row-wise and column-wise, we propose a new algorithm that allows a multi-agent network to perform principal component analysis in a distributed manner. We also verify its validity by numerical experiments. The proposed algorithm is based on the power method for principal component analysis and the average consensus algorithm.

关键词： Signal processing algorithms distributed databases Consensus algorithm Signal processing Partitioning algorithms distributed algorithms Principal component analysis

来源：评论

学校读者我要写书评

暂无评论

Denoising The Wireless Channel Corrupted images Using Machine Learning 20

Denoising The Wireless Channel Corrupted Images Using Machin...

引用

20th IEEE/ACIS International conference on Software Engineering, Artificial Intelligence, Networking and parallel/distributed Computing (SNPD 2019)

作者： Sufiant, Muhammad Khan, Aimal Saced, Takreem Khan, Zirak NUST CEME Dept Comp Engn Islamabad Pakistan

ISBN: (纸本)9781728116518

Digital communication depends on channel coding for the integrity and correct reception of the data. The traditional communication techniques ignore the context and content associated with the received data while mitigating the channel effects. image transmission in digital communication have many major constraints. image quality degrades over wireless channel due to limited characteristic of transmitted data. To mitigate the effects of the noisy channel on the images different image denoising techniques are used such as DCT, Convolutional Denoising Autoencoders, DWT, BM3D, FFT, NLM and Deep Learning. Main objective of this paper is to study the application of machine learning in digital communication to correct errors and remove the effects of channel degradation. This will help us improve the energy consumption, resource utilization, and the Bit Error Rate. It will enable us to communicate with low bandwidth and using minimum resources. Machine learning for denoising the image has attracted substantial attentions because of its high denoising performance. We have constructed a Convolutional Neural Network that will denoise the image which is corrupted by the noisy channel. The existing traditional models only denoise the image for a specific noise type (Gaussian noise) and certain noise level sigma = 25, but our trained network denoise the image for unknown noise level. By using the methods of batch normalization and residual learning the denoising performance is increased. Experiments prove that our trained model displays high efficiency in image denoising. The proposed network is trained in MATLAB and Python with the help of GPU computing which accelerates neural network's performance. The application areas of this paper mainly are in the domains related to digital communications. This include wireless communications, optical communications and line communications. However, as the wireless communications (especially underwater wireless) is extremely error prone,

关键词： Digital Communication Machine Learning Batch Normalization image Transmission Residual Learning image Denoising Deep Learning Convolutional Neural Network

来源：评论

学校读者我要写书评

暂无评论

Lightweight Mobile Devices Indoor Location Based on image Database

Lightweight Mobile Devices Indoor Location Based on Image Da...

引用

International conference on parallel and distributed Systems (ICPADS)

作者： Ran Gao Yanchao Zhao Maoxing Tang College of Computer Science and Technology Nanjing University of Aeronautics and Astronautics China Collaborative Innovation Center of Novel Software Technology and Industrialization China

ISBN: (数字)9781728190747

ISBN: (纸本)9781728183824

Among the numerous indoor localization technologies, image-based solution has great advantages on convenient access from smartphone and its infrastructure-less deployment. However, the image-based localization also suffers from two key disadvantages, which hinders the universal application. Firstly, it requires a large amount of computing and storage resources, which is difficult to achieve for the mobile device, while cloud-based scheme incurs unacceptable delay. Secondly, although this solution doesn't require infrastructure, it still suffers from labor intensive image-database construction and updates. To overcome these limitations, we propose an image-based indoor localization method featured with realtime localization and labor-less image database update. This method mainly innovates in two aspects. First, we propose a mobile device compatible image database compressing framework, which enable realtime and accurate on-device image searching even in a large scenario. Our localization method achieves resource efficiency (in terms of storage and processing) by only keeping image feature vectors, and employing the efficient k-mean tree to search for the best matched image. Secondly, to achieve labor-less image database updating, we mainly add high-quality and informative query image into the database. These query image could compensate the missing information or changed scenario in a up-to-date manner. We conduct real experiments in Android Platform to verify the feasibility and performance of the localization method. Experiment results show that our method has good accuracy (90% location errors are within 1.5m) and high real-time performance (average location delay is less than 0.5s).

关键词： Location awareness Performance evaluation image databases Feature extraction Mobile handsets Real-time systems Delays

来源：评论

学校读者我要写书评

暂无评论

Accelerating Robust Watermarking Through parallelization

Accelerating Robust Watermarking Through Parallelization

引用

International CSI Computer conference, CSICC

作者： Sorour Sheidani Mahmood Fazlali Ziba Eslami Department of Computer Science Shahid Beheshti University G.C. Tehran Iran

ISBN: (数字)9781728159379

ISBN: (纸本)9781728159386

The rapid online usage of digital images has pushed us to accelerate the watermarking method to protect it. Lately, in order to accelerate the applied science and engineering methods multicore architectures are used more and more. Therefore, at this research we have tried to increase the speed of watermarking method on GPU by CUDA. This acceleration by CUDA allows rapid embedding and extraction of the watermarks. For the sake of the fact that better methods have more computational cost, parallelization leads up to the use of a better watermarking method which has a better performance in terms of imperceptibility, robustness, and security. It assists to the rapid secure use of digital images, as well. The results show that the speeds of CUDA implementations are superior to multi cores which have been implemented by OpenMP, although, when the images are big enough OpenMP implementations show a slightly better performance than sequential implementations. The proposed method is evaluated on an NVIDIA 940MX GPU which can reduce the execution time by 50% in comparison with sequential implementations.

关键词： embedded systems graphics processing units image coding image watermarking multiprocessing systems parallel architectures Multiprocessing systems Watermarking parallel architectures digital image watermarking image coding Embedded systems Graphics processing Unit parallelization Digital images acceleration execution time applied science

来源：评论

学校读者我要写书评

暂无评论

distributed matrix methods of compression, masking and noise-resistant image encoding in a high-speed network of information exchange, information processing and aggregation

Distributed matrix methods of compression, masking and noise...

引用

SPIE conference on Future Sensing Technologies

作者： Kapranova, Ekaterina A. Nenashev, Vadim A. Sergeev, Alexander M. Burylev, Dmitry A. Nenashev, Sergey A. St Petersburg State Univ Aerosp Instrumentat 67 Bolshaya Morskaia Str St Petersburg 190000 Russia

ISBN: (数字)9781510631120

ISBN: (纸本)9781510631120

Optical location systems implemented on the basis of high-resolution video cameras are currently used in many areas. These are, for example, medical equipment, traffic control systems, satellite monitoring systems, preventive security systems, object recognition and classification systems, etc. For these systems, the requirements for high-resolution image processing speed of 8K, 16K and more are increasing every year. processing of such information becomes even more difficult when providing a high frequency of reading frames from the matrix of the video camera, especially for systems operating in real time and using high-speed networks of exchange, processing and integration of information. This requires to determine a set of types of information processing procedures: masking, compression, noise-protected coding, etc., for which algorithms should be revised in case of multi-user and multiposition application in distributed information processing and aggregation systems. In this regard, the problems of development and improvement of new ways of representation, compression, storage, masking and error-correcting coding of high-resolution images with a common mathematical basis are relevant. Most information transformation procedures are based on the use of orthogonal bases, in particular orthogonal and quasiorthogonal matrices. The paper presents the results of the search and formation of such bases, the methods of synthesis of quasi-orthogonal matrices for image processing problems that meet the formulated requirements. The methods of guaranteed synthesis of matrices of symmetric, cyclic, block-cyclic and other structures of different orders, assuming economical storage and generation, are proposed. Such matrix bases, which are constantly expanding, provide developers with a wide range of algorithms to choose the most appropriate one from them. The problem of search and study of extreme quasi-orthogonal matrices has great importance for a wider range of information p

关键词： masking images noise-protected coding of high-resolution images compression images information processing and aggregation systems orthogonal bases quasi-orthogonal matrices the problem of search and study of extreme quasi-orthogonal matrices noise-resistant coding of data during their transmission in wireless high-speed networks

来源：评论

学校读者我要写书评

暂无评论

Interactive distributed cloud-based web-server systems for the smart healthcare industry

arXiv

引用

arXiv 2020年

作者： Baurzhanovna, Kondybayeva Almagul National University of Science and Technology MISiS

Subject. This work is dedicated to the questions of the contemporary medical image visualization, the architecture design of the cloud server systems and the using of methods for the .DICOM data representation for the distributed smart healthcare industry ***. In modern medicine and biology, in the pace of research along with the objective need for a constant increase, there is a sharp necessity of the three-dimensional data representation [1]: requiring high-performance methods of the three-dimensional visualization, processing, decomposition, reconstruction and analysis. In many ways, the reason for the sharp growth is associated with the development of the computer technologies (from quantity into quality): the quantitative characteristics of the devices and computers used, the emergence of three-dimensional technologies for processing, visualization and research of data, and, on this basis, the creation of new three-dimensional methods for the human-machine *** methodology. This paper proposes a method for visualization using direct volumetric rendering based on the cubic spline interpolation on a ray emission technology. Modifications of these algorithms, based on a block decomposition, are proposed and still investigated. The server architecture is proposed as a cloud hypervisors server system for the grid processing of the data ***: Data Science for Smart Healthcare, healthcare, internet of things, computer science, interactive visualization, distributed, web-server, cloud, parallel, systems, scientific research, cloud technologies, communications, information, cloud healthcare systems, web-services, web-technologies, information systems, experimental research, DICOM, CT, computed tomography, medical imaging, volume render 2020 MSC: 68-00, 97U99Scope of application. The work aims to investigate the possible contemporary interactive cloud based solutions in the fields of the applied medicine for the smart Healthcare a

关键词： Web services

来源：评论

学校读者我要写书评

暂无评论

parallel multi-scale CNN for image denoising 19

Parallel multi-scale CNN for image denoising

引用

Proceedings of the 5th International conference on Communication and Information processing

作者： Taoling Zhao Yi Zhong Yuanli Wang Wuhan University of Technology Wuhan Hubei

ISBN: (纸本)9781450372589

Deep learning has achieved outstanding performance in the field of image processing. Inspired by the inception model, we propose a deep neural network convolving simultaneously at two different scales. We utilize the convolution kernel of different sizes to carry out the convolution, which can obtain the features of the image at different scales. The features of the output are no longer uniformly distributed, but converge the features with strong correlation. Batch normalization (BN) and residual learning (RL) are utilized to speed up the training process as well as enhance the denoising ability of model. The experimental results of three salt-and-pepper noise levels based on the public dataset BSDS500 show that the newly proposed method can obtain significantly superior performance than current state-of-the-art approaches and higher peak signal to Noise Ratio (PSNR) in salt-and-pepper noise removal.

关键词： feature extraction at different scales residual learning inception model batch normalization image denoising

来源：评论

学校读者我要写书评

暂无评论

Optimization of the processing of images of pollen grains based on the use of their specific characteristics and geometric features

引用

AIP conference Proceedings 2022年第1期2637卷

作者： Isroil Jumanov Rustam Safarov 1Samarkand State University University Boulevard 15 Samarkand 1400000 Uzbekistan

methods have been developed to increase the accuracy of processing of images of pollen grains based on mechanisms for extracting statistical, dynamic, texture and specific characteristics, as well as geometric features of micro-objects. A technique is proposed to increase the accuracy of information processing by the metric characteristics of the contour points of the input and reference images. methods have been developed for point and nonlinear verification of the correspondence of the contours of the input and reference objects based on the mechanisms of isolating, segmenting, interpolating, contrasting, extracting specific characteristics of pollen grains such as frequency, cytoplasm, reticular, spore, texture, morphology of the object and other geometric features of raster images. A set of information processing methods with mechanisms for reducing the zero points of the image contour, reducing the size of rasters, scaling, threshold and level control of dynamic parameters during image recognition and classification, adjusting the points of the color and brightness picture parameters, fixing the initial values, segment centroid and identification are proposed and implemented. Implemented a computer complex of micro objects view in C ++ in the parallel computing environment “CUDA”.

关键词：

来源：评论

学校读者我要写书评

暂无评论

parallel multi-scale CNN for image denoising 5

Parallel multi-scale CNN for image denoising

引用

5th International conference on Communication and Information processing, ICCIP 2019

作者： Zhao, Taoling Zhong, Yi Wang, Yuanli Hubei Key Laboratory of Broadband Wireless Communication and Sensor Networks Wuhan University of Technology WuhanHubei China

ISBN: (纸本)9781450372589

Deep learning has achieved outstanding performance in the field of image processing. Inspired by the inception model, we propose a deep neural network convolving simultaneously at two different scales. We utilize the convolution kernel of different sizes to carry out the convolution, which can obtain the features of the image at different scales. The features of the output are no longer uniformly distributed, but converge the features with strong correlation. Batch normalization (BN) and residual learning (RL) are utilized to speed up the training process as well as enhance the denoising ability of model. The experimental results of three salt- and-pepper noise levels based on the public dataset BSDS500 show that the newly proposed method can obtain significantly superior performance than current state-of-the-art approaches and higher peak signal to Noise Ratio (PSNR) in salt-and-pepper noise removal. © 2019 Association for Computing Machinery.

关键词： Salt and pepper noise

来源：评论

学校读者我要写书评

暂无评论

How to Avoid Zero-Spacing in Fractionally-Strided Convolution? A Hardware-Algorithm Co-Design Methodology

How to Avoid Zero-Spacing in Fractionally-Strided Convolutio...

引用

International conference on High Performance Computing

作者： Yuan Meng Sanmukh Kuppannagari Rajgopal Kannan Viktor Prasanna University of Southern California US Army Research Lab

ISBN: (纸本)9781665410175

Fractionally Strided Convolution (FSC) is a key operation in popular image-based Deep Learning models, for example, back propagation in CNN training, the decoding stage of convolutional auto-encoders and generative CNNs (GAN), etc. FSC typically performs up-convolution on a 2-D grid image, i.e., expands it to a larger one, as compared to conventional (down)-convolution, resulting in more complex computation patterns. Specifically, it introduces additional interleaved zero-spacing (i.e. insertion and padding of zeros) in feature maps that impose excessive computation and memory access overheads on traditional convolution methods such as im2col. The resulting hardware under-utilization is especially severe in layers with large kernels and large strides, commonly seen in typical CNNs and Generative CNNs. In this paper, we propose a methodology to address this challenge using a multi-channel-multi-kernel parallel algorithm, kn2row, to eliminate zero-computations in FSC. We further develop a unified accelerator for kn2row-based convolution and FSC operations in High-Level Synthesis (HLS). Benefiting from the compute-reduction of kn2row, we achieve up to 14.6x improvement in effective resource utilization in typical convolutional auto-decoding layers, GAN layers and backward pass of Nature-CNN, a reinforcement learning bench-marking model. These lead to overall speedup of up to 3.8x in the complete forward or backward propagation phases of the above benchmarks. Our methodology leads up to 8x speedup and 11x better power efficiency than general-purpose processors. Compared with existing GAN accelerators, our methodology achieves higher normalized throughput with high portability.

关键词： Backpropagation Convolution Computational modeling Graphics processing units Benchmark testing Throughput Hardware

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：