检索结果-内蒙古大学图书馆

International Conference on Advanced Computing and Applications (ACOMP)

作者： Trung-Hieu Tran Mammadov, Gasim Sun, Kaicong Simon, Sven Univ Stuttgart Inst Parallel & Distributed Syst Stuttgart Germany

ISBN: (纸本)9781538691861

Light-field imaging has become an emerging technology that brings great benefits to many fields, i.e. in photography, academia, and industry. However, these benefits come with the cost of high computation requirement that limits its applications in practice. This paper presents an accelerated solution for 4D light-field image super-resolution. The acceleration is achieved by the mean of parallel computation using graphics processing units. The selected algorithm is broken into functions which is suitable for parallel execution. Each of the functions is then transformed into GPU kernel and executed at each work-item which is associated with a pixel location in the proposed architecture. Using disparity maps extracted from input 4D light-field as an aid for super-resolution task, the proposed approach can successfully super-resolute an input 4D light-field by the factor of 4 horizontally and vertically. Two strategies, Y-RGB and RGB, are proposed to handle color images. Y-RGB is suitable for high-speed processing constraints while RGB is more preferable if output quality is the main concern. Experimental results show that the proposed approach can achieve the speed up of 203x and 71x compared to CPU implementation for Y-RGB and RGB strategy respectively. Regarding output quality, the proposed approach generates a shaper high-resolution image with more details compared to the baseline methods.

关键词： Spatial resolution Graphics processing units Imaging Estimation Computer architecture Three-dimensional displays

来源：评论

学校读者我要写书评

暂无评论

A Memory Efficient parallel Particle-Based Volume Rendering for Large-Scale distributed Unstructured Volume Datasets in HPC Environments 18th

A Memory Efficient Parallel Particle-Based Volume Rendering ...

引用

18th Annual Asia Simulation Conference (AsiaSim)

作者： Yamaoka, Yoshiaki Hayashi, Kengo Sakamoto, Naohisa Nonaka, Jorji Kobe Univ Kobe Hyogo Japan RIKEN Ctr Computat Sci Kobe Hyogo Japan

ISBN: (纸本)9789811328534;9789811328527

In recent years, the size and complexity of the datasets generated by the large-scale numerical simulations using modern HPC (High Performance Computing) systems have continuously increasing. These generated datasets can possess different formats, types, and attributes. In this work, we have focused on the large-scale distributed unstructured volume datasets, which are still applied on numerical simulations in a variety of scientific and engineering fields. Although volume rendering is one of the most popular techniques for analyzing and exploring a given volume data, in the case of unstructured volume data, the time-consuming visibility sorting becomes problematic as the data size increases. Focusing on an effective volume rendering of large-scale distributed unstructured volume datasets generated in HPC environments, we opted for using the well-known PBVR (Particle-based Volume Rendering) method. Although PBVR does not require any visibility sorting during the rendering process, the CPU-based approach has a notorious image quality and memory consumption tradeoff. This is because that the entire set of the intermediate rendering primitives (particles) was required to be stored a priori to the rendering processing. In order to minimize the high pressure on the memory consumption, we propose a fully parallel PBVR approach, which eliminates the necessity for storing these intermediate rendering primitives, as required by the existing approaches. In the proposed method, each set of the rendering primitives is directly converted to a partial image by the processes, and then they are gathered and merged by the utilized parallel image composition library (234Compositor). We evaluated the memory cost and processing time by using a real CFD simulation result, and we could verify the effectiveness of our proposed method compared to the already existing parallel PBVR method.

关键词： Particle-based Volume Rendering Unstructured volume dataset 234 image composition In-situ visualization

来源：评论

学校读者我要写书评

暂无评论

Fast Nonlocal Diffusion by Stacking Local Filters for image Denoising

Fast Nonlocal Diffusion by Stacking Local Filters for Image ...

引用

2019国际计算机前沿大会

作者： Peng Qiao Weichu Sun Yong Dou Rongchun Li Science and Technology on Parallel and Distributed Laboratory National University of Defense Technology University of South China

Many image denoising methods achieve state-of-the-art performance with little consideration of computation efficiency. With the popularity of high-definition imaging devices, these denoising methods do not scale well to the high-definition images. Therefore, a fast image denoising method is proposed to meet the demand of processing highdefinition images. Based on the analysis of the distribution of the distance distrs between the similar patches and their reference patches from a semantic aspect, the large distrss was found to occur while their contribution to the overall distribution was small. Therefore, the nonlocal filters was replaced in trainable non-local reaction diffusion(TNLRD)with local filters. For image with 4096 × 4096 resolution, the proposed method runs about 6 times faster than TNLRD via a single-thread CPU implementation. And the GPU implementation of the proposed method is about 10 times faster than the CPU implementation. Furthermore,the proposed model achieves competing denoising performance compared with TNLRD in terms of PSNR and SSIM.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Spectral Radius Based Evaluation Function for Automatic Focusing

A Spectral Radius Based Evaluation Function for Automatic Fo...

引用

IEEE International Symposium on parallel and distributed processing with Applications and IEEE International Conference on Ubiquitous Computing and Communications (ISPA/IUCC)

作者： Guojin Peng Meini Lv Kuikui Wang Guangxi Colleges and Universities Key Laboratory of Image Processing and Intelligent Information System Wuzhou University WuZhou china

ISBN: (纸本)9781728152103

Curves generated by traditional evaluating functions contain large amount of local extremums, easily leading to focal-plane misjudgments. We propose a new clarity evaluation method based on spectral radius. The spatial information is converted into frequency domain by Discrete Cosine Transformation, then a sector that contains more than 99 percent of the total spectrum energy is extracted, and the radius of this sector is employed to estimate the degree of clarity of current image. Since the scope of spectral radius is steady and limited to image size, it provides effective reference for selection of lower limit of spectral radius which is utilized to avoid local fluctuation and maintain the unimodality of evaluation curves. The experimental results demonstrate that the spectral radius based evaluation method forms curves with fewer local extremums and better robustness.

关键词： discrete cosine transforms gradient methods image recognition spectral radius discrete cosine transforms gradient methods image recognition Curves Clarity Current Pick image SIZE Sectoring SPECTRAL ENERGY evaluation method

来源：评论

学校读者我要写书评

暂无评论

Scalable deep learning on distributed infrastructures: Challenges, techniques and tools

arXiv

引用

arXiv 2019年

作者： MAYER, RUBEN JACOBSEN, HANS-ARNO Technical University of Munich Boltzmannstr. 3 Garching85748 Germany

Deep Learning (DL) has had an immense success in the recent past, leading to state-of-the-art results in various domains such as image recognition and natural language processing. One of the reasons for this success is the increasing size of DL models and the proliferation of vast amounts of training data being available. To keep on improving the performance of DL, increasing the scalability of DL systems is necessary. In this survey, we perform a broad and thorough investigation on challenges, techniques and tools for scalable DL on distributed infrastructures. This incorporates infrastructures for DL, methods for parallel DL training, multitenant resource scheduling and the management of training and model data. Further, we analyze and compare 11 current open-source DL frameworks and tools and investigate which of the techniques are commonly implemented in practice. Finally, we highlight future research trends in DL systems that deserve further research. Copyright © 2019, The Authors. All rights reserved.

关键词： Natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

A Tone-Mapping Technique Based on Histogram Using a Sensitivity Model of the Human Visual System

引用

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 2018年第4期65卷 3469-3479页

作者： Khan, Ishtiaq Rasool Rahardja, Susanto Khan, Muhammad Murtaza Movania, Muhammad Mobeen Abed, Fidaa King Abdulaziz Univ Fac Comp & IT Jeddah Saudi Arabia Univ Jeddah Fac Comp & IT Jeddah 21589 Saudi Arabia Northwestern Polytech Univ Xian 710072 Shaanxi Peoples R China DHA Suffa Univ Karachi 710072 Pakistan

High-dynamic-range (HDR) images require tone mapping to be displayed properly on lower dynamic range devices. In this paper, a tone-mapping algorithm that uses histogram of luminance to construct a lookup table (LUT) for tone mapping is presented. Characteristics of the human visual system (HVS) are used to give more importance to visually distinguishable intensities while constructing the histogram bins. The method begins with constructing a histogram of the luminance channel, using bins that are perceived to be uniformly spaced by the HVS. Next, a refinement step is used, which removes the pixels from the bins that are indistinguishable by the HVS. Finally, the available display levels are distributed among the bins proportionate to the pixels counts thus giving due consideration to the visual contribution of each bin in the image. Quality assessment using both quantitative evaluations and user studies suggests that the presented algorithm produces tone-mapped images that are visually pleasant and preserve details of the original image better than the existing methods. Finally, implementation details of the algorithm on GPU for parallel processing are presented, which could achieve a significant gain in speed over CPU-based implementation.

关键词： High dynamic range (HDR) human visual system (HVS) image enhancement tone mapping

来源：评论

学校读者我要写书评

暂无评论

Semi-Supervised Generative Adversarial Model for 3D Recognition

Semi-Supervised Generative Adversarial Model for 3D Recognit...

引用

IEEE International Symposium on parallel and distributed processing with Applications and IEEE International Conference on Ubiquitous Computing and Communications (ISPA/IUCC)

作者： Haoqian Wang Shuhao Zhang Yongbing Zhang Zhihong Liu Key Laboratory of Broadband Network & Multimedia Graduate School at Shenzhen Tsinghua University Shenzhen China Shenzhen Institute of Future Technology Shenzhen China

ISBN: (纸本)9781728152103

In this paper, we focus on the issue of 3D point cloud recognition utilizing semi-supervised generative adversarial network, composed of two parts: classification net and generative adversarial net. Using the adversarial criterion, the generator learns to establish the mapping relationship from one-dimensional noise data to 3D point cloud, which can remedy the insufficient data available faced by traditional supervised methods, and the adversarial discriminator acts as a powerful 3D object descriptor to get the point cloud into the right category and eliminate the fake data brought by the generator. In order to fix the rotation problem of point cloud data and improve the recognition performance of the model, we also introduce the spatial transformation networks into the discriminator. Compared with the current supervised and unsupervised methods, the impressive experimental results we show illustrate the effectiveness of the proposed method.

关键词： image classification neural nets object recognition stereo image processing supervised learning Stereo image processing Neural network Semi-supervised Learning image classification object recognition Clouds dynamos Recognition (Psychology)

来源：评论

学校读者我要写书评

暂无评论

Graph Convolutional Networks with Objects for Skeleton-Based Action Recognition

Graph Convolutional Networks with Objects for Skeleton-Based...

引用

IEEE International Symposium on parallel and distributed processing with Applications and IEEE International Conference on Ubiquitous Computing and Communications (ISPA/IUCC)

作者： Xiangbin Shi Haowen Li Fang Liu Deyuan Zhang Jing Bi Zhaokui Li School of Computer Shenyang Aerospace University Shenyang China

ISBN: (纸本)9781728152103

Recently, graph convolutional neural networks has become a research hotspot for skeleton-based action recognition because of its excellent performance on graph structure data. Compared to traditional methods, it can explicitly exploit the natural connectivity among the joints and improve greater expressive power. In this paper, we propose a two-stream graph convolutional networks with objects for skeleton-based action recognition. An algorithm is designed for matching similar skeleton in adjacent frames, so that we can get the right skeletons which belong to the same person. It performs well when there are other irrelevant persons in the scene. In addition, other features are less employed except for the human joint in skeleton-based action recognition. We introduce limbs orientation information and related objects information. The related objects are treated as joint points which link with hands. The two-stream networks are built to model coordinate features and orientation features respectively, the results of two streams are fused to one. We get good results on the Kinetics dataset with our methods.

关键词： convolutional neural nets feature extraction graph theory image motion analysis image recognition image representation learning (artificial intelligence) object recognition video signal processing image motion analysis Video signal processing image representation image recognition Graph theory object recognition Feature extraction Line graph Recognition (Psychology) Joints

来源：评论

学校读者我要写书评

暂无评论

Auto-Resource Provisioning for MapReduce-Based Multiple Object Tracking in Video 18

Auto-Resource Provisioning for MapReduce-Based Multiple Obje...

引用

19th International Conference on distributed Computing and Networking (ICDCN)

作者： Singh, Gurinderbeer Majumdar, Shikharesh Rajan, Sreeraman Carleton Univ Syst & Comp Engn Dept Ottawa ON Canada

ISBN: (纸本)9781450363723

Use of complex image analysis and globally optimal techniques make the current Multiple Object Tracking (MOT) methods for video analysis computationally slow. An important issue in this context is meeting the specific latency requirement for a given application while processing large scale video data. This is especially important in emergency situations such as accidents, natural calamities, and terrorist attacks. This paper introduces a latency reducing MapReduce/Hadoop-based parallel solution for MOT. The system includes an Auto-Resource Provisioning technique for determining the number of Hadoop nodes required to process the MOT job within a user specified deadline. The estimated number of nodes are then provisioned by the system and the MOT application is executed on the Hadoop cluster comprising the desired number of nodes. A prototype is built using the AWS EC2 cloud. A performance analysis is performed using measurements made on the prototype and insights gained into system behavior and performance are presented.

关键词： Video processing Multiple Object Tracking Cloud Computing distributed Systems MapReduce Apache Hadoop

来源：评论

学校读者我要写书评

暂无评论

DP-Pose: Multi-Person Pose Estimation in Video Sequence through Dynamic Programming

DP-Pose: Multi-Person Pose Estimation in Video Sequence thro...

引用

IEEE International Symposium on parallel and distributed processing with Applications and IEEE International Conference on Ubiquitous Computing and Communications (ISPA/IUCC)

作者： Deyuan Zhang Junyuan Wang Xiangbin Shi Zhaokui Li Fang Liu Cuiwei Liu School of Computer Shenyang Aerospace University Shenyang China

ISBN: (纸本)9781728152103

Human pose estimation is an important research topic in the field of computer vision. Bottom-up methods such as OpenPose have become prevalent pose estimation methods because of its high efficiency and detection speed. But for poor quality video, the keypoints obtained by the bottom-up method have suffered from jitter and loss. In this paper, we propose DP-Pose, which constructs candidate pose sequence by selecting points in a region of the heatmap, improves constraint function by combining distance and confidence, and solves the optimal keypoints location by dynamic programming. The experimental results on class dataset show that DP-Pose can recover missing keypoint using heatmap data and obtain stable pose from candidate pose sequences.

关键词： computer vision dynamic programming image sequences pose estimation video signal processing pose estimation dynamic programming Video signal processing images computer vision image sequences Video sequences candidate species

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：