检索结果-内蒙古大学图书馆

3D Ear Shape Matching Using Joint -Entropy

Journal of Computer Science & Technology 2015年第3期30卷 565-577页

作者：孙晓鹏李思慧韩枫魏小鹏 Institute of Computer Systems School of Computer and Information Technology Liaoning Normal University Dalian 116029 China Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia Beijing University of Posts and Telecommunications Beijing 100876 China School of Mechanical and Engineering Dalian University of Technology Dalian 116024 China Key Laboratory of Advanced Design and Intelligent Computing Dalian University Dalian 116622 China

In this article, we investigate the use of joint a-entropy for 3D ear matching by incorporating the local shape feature of 3D ears into the joint a-entropy. First, we extract a sut^cient number of key points from the 3D ear point cloud, and fit the neighborhood of each key point to a single-value quadric surface on product parameter regions. Second, we define the local shape feature vector of each key point as the sampling depth set on the parametric node of the quadric surface. Third, for every pair of gallery ear and probe ear, we construct the minimum spanning tree （MST） on their matched key points. Finally, we minimize the total edge weight of MST to estimate its joint a-entropy the smaller the entropy is, the more similar the ear pair is. We present several examples to demonstrate the advantages of our algorithm, including low time complexity, high recognition rate, and high robustness. To the best of our knowledge, it is the first time that, in computer graphics, the classical information theory of joint a-entropy is used to deal with 3D ear shape recognition.

关键词： joint a-entropy minimum spanning tree local shape feature ear matching ear recognition

来源：评论

学校读者我要写书评

暂无评论

New Region-based Image Fusion Scheme Using the Discrete Wavelet Frame Transform

New Region-based Image Fusion Scheme Using the Discrete Wave...

引用

World Congress on intelligent Control and Automation

作者： Lianhai Wang Junping Du Suguo Zhu Dan Fan JangMyung Lee Beijing Key Laboratory of Intelligent Telecommunication Software and Multimedia School of Computer Science Beijing University of Posts and Telecommunications No.10 Xitucheng Road Haidian District Beijing China Department of Electronics Engineering Pusan National University Busan Korea

ISBN: (纸本)9781467384155

In the field of image fusion, multi-source image fusion methods based on the pixel level can be classified according to two categories: image fusion based on the spatial domain and on the transform domain. When the coefficients of a fusion image are combined, we usually apply a fusion rule based on pixels or windows, whereas the local features in the image are usually not represented. To resolve this problem, we propose a fusion method based on the discrete wavelet frame transform and regional characteristics. First, the transform coefficients are obtained for the two source images using the discrete wavelet frame transform. The average image is acquired after averaging the transform coefficients, which can roughly represent the features of these two source images. The average image is then segmented based on region features and the region coordinates obtained by segmentation are mapped onto the coefficients of the source images obtained by the discrete wavelet frame transform. Finally, the coefficients of each region can be combined using the specific fusion rules. Our experimental results demonstrate that the proposed method performs better compared with the fusion method based on the Laplacian pyramid transform and that based on the shiftShift-invariant discrete wavelet transform at preserving the regional features of source images, as well as delivering better performance in terms of both visuals effects and an objective index.

关键词： Image fusion Discrete wavelet transforms Fuses software Multimedia communication Computer science image fusion source image discrete wavelet transform Fuses Multimedia communications fusion method computer science Transform computer software wavelet transforms wavelets Laplacian pyramid Frame

来源：评论

学校读者我要写书评

暂无评论

Neural Incremental Attribute Learning Based on Principal Component Analysis

Neural Incremental Attribute Learning Based on Principal Com...

引用

2016 IEEE International Conference on Big Data Analysis

作者： Ting Wang Wei Zhou Xiaoyan Zhu Fangzhou Liu Sheng-Uei Guan State Key Laboratory of Intelligent Technology and Systems Tsinghua University Wuxi Research Institute of Applied Technologies Tsinghua University Dept. of Computer Science University of Liverpool Dept. of Computer Science and Software Engineering Xi'an Jiaotong-Liverpool University

ISBN: (纸本)9781467395922

Feature Extraction(FE) based on Principal Component Analysis(PCA) can effectively improve classification results by reducing the interference among features. However, such a good method has not been employed in previous studies of Incremental Attribute Learning(IAL), a novel machine learning strategy, where features are gradually trained one by one in order to remove interference among features and improve classification results. This study proposed a preprocessing for neural IAL algorithm based on feature extraction with PCA. Experimental results show that this approach is not only very efficient, but also applicable for pattern classification performance improvement.

关键词： pattern Classification Feature Extraction Principal Component Analysis Neural Networks Incremental Attribute Learning

来源：评论

学校读者我要写书评

暂无评论

Runtime model based approach to IoT application development

引用

Frontiers of Computer Science 2015年第4期9卷 540-553页

作者： Xing CHEN College of Mathematics and Computer Science Fuzhou University Fuzhou 350116 China Fujian Provincial Key Laboratory of Networking Computing and Intelligent Information Processing Fuzhou 350116 China Key Laboratory of High Confidence Software Technologies (Ministry of Education) Beijing 100871 China School of Electronics Engineering and Computer Science Peking University Beijing 100871 China

The internet of things （loT） attracts great interest in many application domains concerned with monitoring and ：ontrol of physical phenomena. However, application devel- opment is still one of the main hurdles to a wide adoption of IoT technology. Application development is done at a low level, very close to the operating system and requires pro- grammers to focus on low-level system issues. The under- lying APIs can be very complicated and the amount of data collected can be huge. This can be very hard to deal with as a developer. In this paper, we present a runtime model based approach to IoT application development. First, the manage- ability of sensor devices is abstracted as runtime models that are automatically connected with the corresponding systems. Second, a customized model is constructed according to a personalized application scenario and the synchronization be- tween the customized model and sensor device runtime mod- els is ensured through model transformation. Thus, all the application logic can be carried out by executing programs on the customized model. An experiment on a real-world ap- plication scenario demonstrates the feasibility, effectiveness, and benefits of the new approach to IoT application develop- ment.

关键词： IoT application development software architec-ture models at runtime

来源：评论

学校读者我要写书评

暂无评论

Denoising Auto-Encoders toward Robust Unsupervised Feature Representation

Denoising Auto-Encoders toward Robust Unsupervised Feature R...

引用

International Joint Conference on Neural Networks

作者： Wei Xiong Bo Du Lefei Zhang Liangpei Zhang Dacheng Tao State Key Laboratory of Software Engineering School of Computer Science Wuhan University China Department of Computing The Hong Kong Polytechnic University Kowloon Hong Kong State Key Laboratory of Information Engineering in Surveying Mapping and Remote Sensing Wuhan University China Centre for Quantum Computation & Intelligent Systems University of Technology Sydney NSW 2007 Australia

ISBN: (纸本)9781509006212

Deep networks like the convolutional neural network and its variants usually learn hierarchical features from labeled images, which is very expensive to obtain. How can we find an unsupervised way to effectively extract deep and abstract features from images without annotations? Even from large qualities of images with noise? In this paper, we propose a robust deep neural network, named as stacked convolutional denoising auto-encoders (SCDAE), which can map raw images to hierarchical representations in an unsupervised manner. Our network is elaborately designed to fit for the visual recognition tasks. It is established by stacking the denoising auto-encoders. Unlike the prior works, in the training phase, the auto-encoders are trained patch-wisely so that the latent features can be applied to powerful regularizers for better representation;in the inference phase, the denoising auto-encoders are stacked convolutionally, hence the generated feature maps in the higher layers can preserve the coherent structures within the features in the lower layers. To achieve better performance, we apply whitening to each layer to sphere the input features. Our network is evaluated on the challenging image datasets MNIST, CIFAR-10 and STL-10 and demonstrates superior performance to the state-of-the-art unsupervised networks.

关键词： Noise reduction Convolutional codes Training Feature extraction Kernel Optimized production technology Robustness

来源：评论

学校读者我要写书评

暂无评论

On Secrecy Performance of MISO SWIPT Systems with TAS and Imperfect CSI

On Secrecy Performance of MISO SWIPT Systems with TAS and Im...

引用

作者： Pan, Gaofeng Lei, Hongjiang Deng, Yansha Fan, Lisheng Yang, Jing Chen, Yunfei Ding, Zhiguo Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing School of Electronic and Information Engineering Southwest University Chongqing400715 China School of Computing and Communications Lancaster University LancasterLA14YW United Kingdom Chongqing Key Laboratory of Mobile Communications Technology Chongqing University of Posts and Telecommunications Chongqing400065 China Centre for Telecommunications King's College London LondonWC2R2LS United Kingdom School of Computer Science and Educational Software Guangzhou University Guangzhou510000 China Department of Electronic Engineering Shantou University Guangdong515063 China School of Information Engineering Yangzhou University Yangzhou225009 China School of Engineering University of Warwick CoventryCV47AL United Kingdom

In this paper, a multiple-input single-output (MISO) simultaneous wireless information and power transfer (SWIPT) system, including one base station (BS) equipped with multiple antennas, one desired single-antenna information receiver (IR), and N (N > 1) single-antenna energy-harvesting receivers (ERs) is considered. Assuming that the information signal to the desired IR may be eavesdropped by ERs if ERs are malicious, we investigate the secrecy performance of the target MISO SWIPT system when imperfect channel state information (CSI) is available and adopted for transmit antenna selection at the BS. Considering that each eavesdropping link experiences independent but not necessarily identically distributed Rayleigh fading, the closed-form expressions for the exact and the asymptotic secrecy outage probability, and the average secrecy capacity are derived and verified by simulations. Furthermore, the optimal power splitting factor is derived for each ER to realize the tradeoff between the energy harvesting and the information eavesdropping. Our results reveal the impact of the imperfect CSI on the secrecy performance of MISO SWIPT systems in the presence of multiple wiretap channels. © 2016 IEEE.

关键词： Energy harvesting

来源：评论

学校读者我要写书评

暂无评论

Situation Awareness and Emergency Response Using Cross-domain Data in Typhoon Haiyan

Situation Awareness and Emergency Response Using Cross-domai...

引用

ISSAT International Conference on Modeling of Complex Systems and Environments, MCSE 2015

作者： Deng, Qing Deng, Xiaolong Liu, Yi Zhang, Hui Department of Engineering Physics Tsinghua University Beijing China Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia Beijing University of Posts and Telecommunications Beijing China

Crisis response is involved in a complex system consisting of cyber, physics and society domains. Different domains are closely coupled and dynamically evolved. Cross domain analysis is applied for better response of natural disasters. A data-model integration method is proposed to support situation awareness and emergency response planning. Models are used for the prediction of crisis scales and its impact on infrastructure system and local society. Data mining is used to sense society readiness and mentality of affected citizen. The integration of cross-domain data mining and emergency modeling is essential for supporting scientific decision-making (planning and response). The response of Typhoon Haiyan is divided into two periods. Before landfall, physical data and models are important for situation awareness and response plan. After landfall, information is needed for damage assessment and rescue plan. For example, based on word frequency analysis and comparative analysis in Hainan and Guangxi, it was clear that damage in Hainan was much severe than that in Guangxi after Typhoon Haiyan landed in China. This study contributes to effective crisis response.

关键词： Big data

来源：评论

学校读者我要写书评

暂无评论

ACCELERATE CONVOLUTIONAL NEURAL NETWORKS FOR BINARY CLASSIFICATION VIA CASCADING COST-SENSITIVE FEATURE

ACCELERATE CONVOLUTIONAL NEURAL NETWORKS FOR BINARY CLASSIFI...

引用

IEEE International Conference on Image Processing

作者： Junbiao Pang Huihuang Lin Li Su Chunjie Zhang Weigang Zhang Lijuan Duan Qingming Huang Baocai Yin Beijing Key Laboratory of Multimedia and Intelligent Software Technology China School of Computer and Control Engineering University of Chinese Academy of Sciences China School of Computer Science and Technology Harbin Institute of Technology at Weihai China College of Computer Science and Technology Beijing University of Technology China

ISBN: (纸本)9781467399623

Convolutional Neural Networks (CNNs) have delivered impressive state-of-the-art performances for many vision tasks, while the computation costs of these networks during test-time are notorious. Empirical results have discovered that CNNs have learned the redundant representations both within and across different layers. When CNNs are applied for binary classification, we investigate a method to exploit this redundancy across layers, and construct a cascade of classifiers which explicitly balances classification accuracy and hierarchical feature extraction costs. Our method cost-sensitively selects feature points across several layers from trained networks and embeds non-expensive yet discriminative features into a cascade. Experiments on binary classification demonstrate that our framework leads to drastic test-time improvements, e.g., possible 47.2× speedup for TRECVID upper body detection, 2.82× speedup for Pascal VOC2007 People detection, 3.72× for INRIA Person detection with less than 0.5% drop in accuracies of the original networks.

关键词： Neural networks Convolution Acceleration Optimization Upper bound Computers Computer science

来源：评论

学校读者我要写书评

暂无评论

Robust control of flat systems using sliding mode approach

Robust control of flat systems using sliding mode approach

引用

American Control Conference

作者： Liming Chen Yingmin Jia Junping Du Seventh Research Division and the Center for Information and Control School of Automation Science and Electrical Engineering Beihang University (BUAA) Beijing 100191 China Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia School of Computer Science and Technology Beijing University of Posts and Telecommunications 100876 China

ISBN: (纸本)9781467386838

In this paper, sliding mode approach is used to deal with the robust control of flat systems. Model of flat systems with disturbance and parametric uncertainty is established. A scheme to design sliding mode control laws is proposed to guarantee the robustness of the systems. For the case that the bounds of disturbance and parametric uncertainty are unknown, adaptive controllers are designed to estimate the bounds or unknown parameters. The stability is proved with Lyapunov method. Simulations of a magnetic levitation system are given to illustrate the effectiveness of the approach.

关键词： Sliding mode control Uncertainty Tracking Lyapunov methods Robustness Feedforward neural networks Adaptive control feedforward neural nets Variable structure systems Lyapunov methods Robust control adaptive control Magnetic levitation systems tracking Uncertainty Robustness Sliding mode FLAT adaptive controller disturbance

来源：评论

学校读者我要写书评

暂无评论

Sequential Bag-of-Words model for human action classification

引用

CAAI Transactions on Intelligence Technology 2016年第2期1卷 125-136页

作者： Hong Liu Hao Tang Wei Xiao ZiYi Guo Lu Tian Yuan Gao School of Electronic and Computer Engineering Peking University Shenzhen Graduate Shcool 518055 China Engineering Lab on Intelligent Perception for Internet of Things (ELIP) Peking University Shenzhen Graduate School 518055 China Key Laboratory of Machine Perception Peking University 100871 China School of Software and Microelectronics Peking University 100871 China Institute of Computer Science Christian-Albrechts-University 24118 Germany

Recently, approaches utilizing spatial-temporal features to form Bag-of-Words （BoWs） models have achieved great success due to their simplicity and effectiveness. But they still have difficulties when distinguishing between actions with high inter-ambiguity. The main reason is that they describe actions by orderless bag of features, and ignore the spatial and temporal structure information of visual words. In order to improve classification performance, we present a novel approach called sequential Bag-of-Words. It captures temporal sequential structure by segmenting the entire action into sub-actions. Meanwhile, we pay more attention to the distinguishing parts of an action by classifying sub- actions separately, which is then employed to vote for the final result. Extensive experiments are conducted on challenging datasets and real scenes to evaluate our method. Concretely, we compare our results to some state-of-the-art classification approaches and confirm the advantages of our approach to distinguish similar actions. Results show that our approach is robust and outperforms most existing BoWs based classification approaches, especially on complex datasets with interactive activities, cluttered backgrounds and inter-class action ambiguities.

关键词： Action classification Sequential Bag-of-Words STIP Probalibity

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：