检索结果-内蒙古大学图书馆

A Robust Residual Dense neural Network For Countering Antiforensic Attack on Median Filtered images

IEEE signal processing LETTERS 2019年第8期26卷 1132-1136页

作者： Tariang, Diangarti Bhalang Chakraborty, Rajat Subhra Naskar, Ruchira Indian Inst Technol Dept Comp Sci & Engn Kharagpur 721302 W Bengal India Indian Inst Engn Sci & Technol Shibpur Dept Informat Technol Howrah 711103 India

Recently, antiforensic methods have been proposed that invalidate most of the state-of-the-art median filter digital image forensic techniques. Also, the existing counter antiforensic methods decline noticeably when evaluated on small-sized patches in JPEG compressed images. In this letter, we have developed a robust residual dense (neural) network-based counter antiforensic median filter detection technique that exploits local dense connection and residual learning of features for improved classification of images. Experimental results demonstrate that the proposed approach achieves superior performance to state-of-the-art techniques in detecting forgeries, even in small patches, *** compressed images, for both median filtered and antiforensic median filtered images.

关键词： Antiforensic attack deep neural network (DNN) digital image forensics median filter forensics residual dense network (RDN)

来源：评论

学校读者我要写书评

暂无评论

Feature fusion via dual-resolution compressive measurement matrix analysis for spectral image classification

引用

signal processing-image COMMUNICATION 2021年 90卷 116014-116014页

作者： Marcos Ramirez, Juan . Martinez Torre, Jose Ignacio Arguello, Henry Univ Rey Juan Carlos Comp Sci Dept Mostoles Spain Univ Ind Santander Comp Sci Dept Bucaramanga Colombia

In the compressive spectral imaging (CSI) framework, different architectures have been proposed to recover high-resolution spectral images from compressive measurements. Since CSI architectures compactly capture the relevant information of the spectral image, various methods that extract classification features from compressive samples have been recently proposed. However, these techniques require a feature extraction procedure that reorders measurements using the information embedded in the coded aperture patterns. In this paper, a method that fuses features directly from dual-resolution compressive measurements is proposed for spectral image classification. More precisely, the fusion method is formulated as an inverse problem that estimates high-spatial-resolution and low-dimensional feature bands from compressive measurements. To this end, the decimation matrices that describe the compressive measurements as degraded versions of the fused features are mathematically modeled using the information embedded in the coded aperture patterns. Furthermore, we include both a sparsity-promoting and a total-variation (TV) regularization terms to the fusion problem in order to consider the correlations between neighbor pixels, and therefore, improve the accuracy of pixel-based classifiers. To solve the fusion problem, we describe an algorithm based on the accelerated variant of the alternating direction method of multipliers (accelerated-ADMM). Additionally, a classification approach that includes the developed fusion method and a multilayer neural network is introduced. Finally, the proposed approach is evaluated on three remote sensing spectral images and a set of compressive measurements captured in the laboratory. Extensive simulations show that the proposed classification approach outperforms other approaches under various performance metrics.

关键词： Compressive spectral imaging Dual-resolution acquisition systems Feature fusion Spectral image classification

来源：评论

学校读者我要写书评

暂无评论

A method of forest-fire image recognition based on adaboost-BP algorithm

引用

International Journal of Circuits, Systems and signal processing 2019年 13卷 312-319页

作者： Ding, Dehong Zhu, Sisi College of Compter Hunan University of Arts and Science Changde415000 China College of Electrical and Information Engineering Hunan University Changsha410082 China Institute of Mechanical and Electrical Engineering Hezhou University HezhouGuanxi542899 China

In view of the shortcomings of the existing forest flame recognition technology, in order to further improve and perfect some problems of the flame image recognition technology, provide guidance and technical support for fire detection products. The research is focus on flame information of physical characteristics, which combines with image processing techniques and comes up to a recognition way of forest flame that character of highlight area based on image and a flame recognition method of BP neural network for forest fire based on AdaBoost. AdaBoost-BP method uses the BP neural network as the basic weak classifier, and then updates it through the specific algorithm, and constructs a combination of strong classifier. Those methods are mainly extracting Color, Area and Circularity of pretreatment image, then output AdaBoost-BP neural network learning to realize firework recognition. Using single algorithms and single characteristic to analyze, easily led to accuracy is not high. It not only can make up for that shortcoming but also eliminate interference interruption, and improves efficiency of image recognition for *** recognition rate is about 94%, and the training time is about 250 times to achieve convergence. The computation speed is only 0.01292 seconds. © 2019, North Atlantic University Union. All rights reserved.

关键词： Adaptive boosting

来源：评论

学校读者我要写书评

暂无评论

Real-Time Emotion Recognition Framework Based on Convolution neural Network 15th

Real-Time Emotion Recognition Framework Based on Convolution...

引用

15th International Conference on Intelligent Information Hiding and Multimedia signal processing, IIH-MSP 2019, held in conjunction with the 12th International Conference on Frontiers of Information Technology, Applications and Tools, FITAT 2019

作者： Yang, Hanting Zhao, Guangzhe Zhang, Lei Zhu, Na He, Yanqing Zhao, Chunxiao Beijing University of Civil Engineering and Architecture Beijing100000 China

ISBN: (纸本)9789811397097

Efficient emotional state analyzing will enable machines to understand human better and facilitate the development of applications which involve human–machine interaction. Recently, deep learning methods become popular due to their generalization ability, but the disadvantage of complicated computation could not meet the requirements of real-time characteristics. This paper proposes an emotion recognition framework based on convolution neural network, which contains less number of parameters comparatively. In order to verify the proposed framework, we train a network on a large number of facial expression images and then use the pretrained model to predict image frame taken from a single camera. The experiment shows that compared to VGG13, our network reduces the parameters by 147 times. © Springer Nature Singapore Pte Ltd. 2020.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Improving the Efficiency of the Method of stochastic Gradient Identification of Objects in Binary and Grayscale images Using Their Preprocessing

Improving the Efficiency of the Method of Stochastic Gradien...

引用

Digital signal processing and its Applications (DSPA), International Conference on

作者： Magdeev Radik Gilfanovich Tashlinskii Alexander Grigorevich Ulyanovsk State Technical University Ulyanovsk Russia

ISBN: (数字)9781728175287

ISBN: (纸本)9781728175294

In this paper, we consider ways to improve the stochastic gradient method efficiency of object identification for binary and grayscale images using methods of image preprocessing. Identification of an object is understood as the recognition of an object on the image with its parameters estimation. Low-pass filtering and image equalization are considered as preliminary processing. The identification parameters convergence rate is investigated. The optimal sizes of Gaussian filter mask for binary and grayscale images were found based on COIL-20 images.

关键词： Gray-scale Object recognition Low pass filters Convergence Brightness stochastic processes

来源：评论

学校读者我要写书评

暂无评论

Research on an Two-Channel ACNN-LSTM Model for Financial Text Sentiment Analysis

Research on an Two-Channel ACNN-LSTM Model for Financial Tex...

引用

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

作者： Hanxiao Shi Liqiang You Mimi Ren Xiaojun Li School of Management and E-Business Zhejiang Gongshang University Hangzhou China School of Science and Technology Zhejiang Gongshang University Hangzhou College of Commerce Hangzhou China

ISBN: (纸本)9781665426565

This paper proposes a sentiment analysis model based on two-channel attention-driven convolutional neural networks and long short term memory neural networks for financial text. Firstly, this paper uses two different word vector initialization methods to construct classification model by selecting different feature representations and taking full account of the relationship between words. Secondly, this paper adds Attention mechanism based on the context structure to analyze the text to obtain more hidden information. Finally, the experimental results show that our approach is feasible and effective.

关键词： Support vector machines Analytical models Sentiment analysis Computational modeling Redundancy neural networks Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Improved Segmentation of the Intracranial and Ventricular Volumes in Populations with Cerebrovascular Lesions and Atrophy Using 3D CNNs

引用

NEUROINFORMATICS 2021年第4期19卷 597-618页

作者： Ntiri, Emmanuel E. Holmes, Melissa F. Forooshani, Parisa M. Ramirez, Joel Gao, Fuqiang Ozzoude, Miracle Adamo, Sabrina Scott, Christopher J. M. Dowlatshahi, Dar Lawrence-Dewar, Jane M. Kwan, Donna Lang, Anthony E. Symons, Sean Bartha, Robert Strother, Stephen Tardif, Jean-Claude Masellis, Mario Swartz, Richard H. Moody, Alan Black, Sandra E. Goubran, Maged Univ Toronto Sunnybrook Res Inst Hurvitz Brain Sci Program Toronto ON Canada Univ Ottawa Ottawa Hosp Dept Med Fac Med Ottawa ON Canada Thunder Bay Reg Hlth Res Inst Thunder Bay ON Canada York Univ Fac Hlth Dept Psychol Toronto ON Canada Toronto Western Hosp Edmond J Safra Program Parkinsons Dis Toronto ON Canada Toronto Western Hosp Morton & Gloria Shulman Movement Disorders Clin Toronto ON Canada Univ Toronto Dept Med Neurol Div Toronto ON Canada Univ Toronto Dept Med Imaging Toronto ON Canada Univ Western Ontario Robarts Res Inst Schulich Sch Med & Dent Dept Med Biophys London ON Canada Univ Toronto Dept Med Biophys Toronto ON Canada Univ Montreal Montreal Heart Inst Montreal PQ Canada Heart & Stroke Fdn Canadian Partnership Stroke Recovery Toronto ON Canada

Successful segmentation of the total intracranial vault (ICV) and ventricles is of critical importance when studying neurodegeneration through neuroimaging. We present iCVMapper and VentMapper, robust algorithms that use a convolutional neural network (CNN) to segment the ICV and ventricles from both single and multi-contrast MRI data. Our models were trained on a large dataset from two multi-site studies (N = 528 subjects for ICV, N = 501 for ventricular segmentation) consisting of older adults with varying degrees of cerebrovascular lesions and atrophy, which pose significant challenges for most segmentation approaches. The models were tested on 238 participants, including subjects with vascular cognitive impairment and high white matter hyperintensity burden. Two of the three test sets came from studies not used in the training dataset. We assessed our algorithms relative to four state-of-the-art ICV extraction methods (MONSTR, BET, Deep Extraction, FreeSurfer, DeepMedic), as well as two ventricular segmentation tools (FreeSurfer, DeepMedic). Our multi-contrast models outperformed other methods across many of the evaluation metrics, with average Dice coefficients of 0.98 and 0.96 for ICV and ventricular segmentation respectively. Both models were also the most time efficient, segmenting the structures in orders of magnitude faster than some of the other available methods. Our networks showed an increased accuracy with the use of a conditional random field (CRF) as a post-processing step. We further validated both segmentation models, highlighting their robustness to images with lower resolution and signal-to-noise ratio, compared to tested techniques. The pipeline and models are available at: and to enable further investigation of the roles of ICV and ventricles in relation to normal aging and neurodegeneration in large multi-site studies.

关键词： Total intracranial volume Ventricles image segmentation Deep learning Vascular lesions Brain atrophy

来源：评论

学校读者我要写书评

暂无评论

A RD-T Network for Hand Gesture Recognition Based on Millimeter-Wave Sensor

A RD-T Network for Hand Gesture Recognition Based on Millime...

引用

IEEE International Conference on signal and image processing (ICSIP)

作者： Rongjie Huang Zecheng Li Song Wang Ronghao Wang Junjie Li Zhiwei Xu Zhejiang University Hangzhou China Institute of Marine Electronic Engineering Ocean College Zhejiang University Zhoushan China

The following topics are dealt with: feature extraction; learning (artificial intelligence); neural nets; convolutional neural nets; object detection; image segmentation; iterative methods; image classification; medic... 详细信息

ISBN: (数字)9781728168968

ISBN: (纸本)9781728168975

关键词： image recognition Gesture recognition Millimeter wave radar Radar imaging Vision sensors Feature extraction Reliability

来源：评论

学校读者我要写书评

暂无评论

Multiscale noise suppression and feature frequency extraction in SSVEP based on underdamped second-order stochastic resonance

引用

JOURNAL OF neural ENGINEERING 2019年第3期16卷 036032-036032页

作者： Yao, Pulin Xu, Guanghua Jia, Linshan Duan, Jinjin Han, Chengcheng Tao, Tangfei Wang, Yi Zhang, Sicong Xi An Jiao Tong Univ Sch Mech Engn Xian 710049 Shaanxi Peoples R China Xi An Jiao Tong Univ State Key Lab Mfg Syst Engn Xian 710049 Shaanxi Peoples R China Chongqing Univ Sch Mech Engn Chongqing 400044 Peoples R China

Objective. As one of the commonly used control signals of brain-computer interface (BCD, steady-state visual evoked potential (SSVEP) exhibits advantages of stability, periodicity and minimal training requirements. However, SSVEP retains the non-linear, non-stationary and low signal-to-noise ratio (SNR) characteristics of EEG. The traditional SSVEP extraction methods regard noise as harmful information and highlight the useful signal by suppressing the noise. In the collected EEG, noise and SSVEP are usually coupled together, the useful signal is inevitably attenuated while the noise is suppressed. Also, an additional band-pass filter is needed to eliminate the multi-scale noise, which causes the edge effect. Approach. To address this issue, a novel method based on underdamped second-order stochastic resonance (USSR) is proposed in this paper for SSVEP extraction. Main results. A synergistic effect produced by noise, useful signal and the nonlinear system can force the energy of noise to be transferred into SSVEP, and hence amplifying the useful signal while suppressing multi-scale noise. The recognition performances of detection are compared with the widely-used canonical coefficient analysis (CCA) and multivariate synchronization index (MSI). Significance. The comparison results indicate that USSR exhibits increased accuracy and faster processing speed, which effectively improves the information transmission rate (ITR) of SSVEP-based BCI.

关键词： brain-computer interface steady-state visual evoked potential canonical coefficient analysis underdamped second-order stochastic resonance multivariate synchronization index

来源：评论

学校读者我要写书评

暂无评论

Modelling Mobile signal Strength by Combining Geospatial Big Data and Artificial Intelligence 2020

Modelling Mobile Signal Strength by Combining Geospatial Big...

引用

Proceedings of the 2020 4th International Conference on Vision, image and signal processing

作者： Paolo Fraccaro Matt Benatan Katharina Reusch Clyde Fare Blair Edwards Edward Pyzer-Knapp IBM Research Europe Hartree Centre Daresbury UK

ISBN: (纸本)9781450389532

Estimating mobile signal strength accurately is a crucial task for network providers and their customers. However, current methodologies to estimate mobile signal strength present limitations in their practical implementation (i.e. physical based models), portability (i.e. spatial interpolation methods), simplicity and accuracy (i.e. path loss models). In this paper we present a novel approach that takes advantage of geospatial Big Data and advanced Artificial Intelligence to predict mobile signal strength at scale. Particularly, we used open access geo-spatial information about weather, tree coverage, land use, imperviousness, altitude and network infrastructure (i.e. a total of 174 features) to train and test uncertainty-aware artificial neural networks to predict mobile signal strength on data from the NetBravo crowdsourcing platform across all the United Kingdom (UK). Our model scored a best performance of 7.9 (standard deviation of 0.2) dBm for Root Mean Squared Error and 5.7 (standard deviation of 0.4) dBm for the Mean Absolute Error. Feature importance analysis showed that mobile cell tower characteristics and geospatial features showing the distribution of imperviousness and tree cover density over the line of sight between the mobile cell tower and the receiver as well as relative humidity were among the top 20 most important features.

关键词： Mobile signal Strength Artificial neural Networks Learning Big Data Geospatial

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：