检索结果-内蒙古大学图书馆

Towards facial micro-expression detection and classification using modified multimodal ensemble learning approach

INFORMATION FUSION 2025年 115卷

作者： Zhang, Fuli Liu, Yu Yu, Xiaoling Wang, Zhichen Zhang, Qi Wang, Jing Zhang, Qionghua Hunan Univ Informat Technol Informat Technol Res Inst Changsha 410151 Hunan Peoples R China China Med Univ Sch Forens Med 77 Puhe RdNorth New Area Shenyang 110122 Liaoning Peoples R China Coll Heilongjiang Natl Training Vocat Educ Harbin 152302 Heilongjiang Peoples R China Shenyang Urban Construct Univ Coll Arts & Media Shenyang 110167 Liaoning Peoples R China Shenyang Univ Technol Sch Informat Sci & Engn Shenyang 110870 Liaoning Peoples R China

A micro-expression is a fleeting, delicate and localized facial gesture. It can expose the true feelings that someone is trying to hide and is seen to be a crucial indicator for spotting lies. Because of its possible applications in a variety of sectors, micro-expression research has garnered a lot of attention. The accuracy of micro-expression recognition still needs to be improved, though, because of the brief and weak motions that make up micro- expressions. In recent years, Deep convolution neural methods have depicted a higher degree of efficiency for complex challenge of face detection. Although several attempts were made for micro-expression recognition (MER), the problem is far from being resolved problem which is portrayed by the lowest accuracy rate depicted by the other models. In this study, present a Facial Micro-Expression Detection and Classification using Modified Multimodal Ensemble Learning (FMEDC-MMEL) approach. The major intention of the FMEDC-MMEL technique lies in the proficient identification of MEs that exist in the facial images. As a pre-processing phase, the FMEDCMMEL technique exploits histogram equalization (HE) approach to improve the contrast level of the image. In the FMEDC-MMEL technique, improved densely connected networks (DenseNet) model is used for learning feature patterns from the pre-processed images. To enhance the proficiency of the improved DenseNet model, stochastic gradient descent (SGD) approach is used for hyperparameter selection process. For facial ME detection, the FMEDC-MMEL technique follows an ensemble of three classifiers namely bi-directional gated recurrent unit (Bi-GRU), long short-term memory (LSTM) and extreme learning machine (ELM). A tailored ensemble learning approach is shown, which combines many machine learning models to improve classification performance and detection accuracy. Sophisticated feature extraction methods are utilized to extract the subtle aspects of micro-expressions, and precision is ma

关键词： Micro-expression detection Facial image Multimodal Ensemble learning stochastic Gradient Descent

来源：评论

学校读者我要写书评

暂无评论

Pattern-invariant Unrolling for Robust Demosaicking 32

Pattern-invariant Unrolling for Robust Demosaicking

引用

32nd European signal processing Conference (EUSIPCO)

作者： Muller, Matthieu Picone, Daniele Mura, Mauro Dalla Ulfarsson, Magnus O. Univ Grenoble Alpes CNRS Grenoble INP GIPSA Lab F-38000 Grenoble France IUF F-38000 Grenoble France Univ Iceland Fac Elect & Comp Engn IS-101 Reykjavik Iceland

ISBN: (纸本)9789464593617;9798331519773

To acquire color images, most commercial cameras rely on color filter arrays (CFAs), which are a pattern of color filters overlaid over the sensor's focal plane. Demosaicking describes the processing techniques to reconstruct a full color image for all pixels on the focal plane array. Most demosaicking methods are tailored for a specific CFA, and tend to work poorly for others. In this work we present an algorithm for demosaicking a wide variety of CFAs. The proposed method allows to blend the knowledge of the CFA with information coming from data, employing a novel transformation and pattern-invariant loss function. The method is based on the unrolling of an algorithm based on a neural network learned on available examples. Preliminary experiments over RGB and RGBW CFAs show that the method performs well over a range of CFAs and is competitive for CFAs for which competing methods were tailored to work well on.

关键词： Demosaicking unrolling image processing deep learning color filter arrays

来源：评论

学校读者我要写书评

暂无评论

IMPLICIT neural MULTIPLE DESCRIPTION FOR DNA-BASED DATA STORAGE 49

IMPLICIT NEURAL MULTIPLE DESCRIPTION FOR DNA-BASED DATA STOR...

引用

49th IEEE International Conference on Acoustics, Speech, and signal processing (ICASSP)

作者： Le, Trung Hieu Pic, Xavier Mateos, Jeremy Antonini, Marc Cote dAzur Univ I3S Lab Sophia Antipolis France CNRS UMR 7271 Sophia Antipolis France

ISBN: (纸本)9798350344868;9798350344851

DNA exhibits remarkable potential as a data storage solution due to its impressive storage density and long-term stability, stemming from its inherent biomolecular structure. However, developing this novel medium comes with its own set of challenges, particularly in addressing errors arising from storage and biological manipulations. These challenges are further conditioned by the structural constraints of DNA sequences and cost considerations. In response to these limitations, we have pioneered a novel compression scheme and a cutting-edge Multiple Description Coding (MDC) technique utilizing neural networks for DNA data storage. Our MDC method introduces an innovative approach to encoding data into DNA, specifically designed to withstand errors effectively. Notably, our new compression scheme overperforms classic image compression methods for DNA-data storage. Furthermore, our approach exhibits superiority over conventional MDC methods reliant on auto-encoders. Its distinctive strengths lie in its ability to bypass the need for extensive model training and its enhanced adaptability for fine-tuning redundancy levels. Experimental results demonstrate that our solution competes favorably with the latest DNA data storage methods in the field, offering superior compression rates and robust noise resilience.

关键词： DNA data storage Multiple Description Coding (MDC) Implicit neural Network (INR) Quaternary Shannon Fano Entropy Coder (SFC4)

来源：评论

学校读者我要写书评

暂无评论

Differential CNN and KELM integration for accurate liver cancer detection

引用

BIOMEDICAL signal processing AND CONTROL 2024年 95卷

作者： Jesi, P. Maria Daniel, V. Antony Asir Loyola Inst Technol & Sci Comp Sci & Engn Kanyakumari Tamilnadu India Loyola Inst Technol & Sci Elect & Commun Engn Kanyakumari Tamilnadu India

Liver cancer is a significant global health concern, with its prevalence steadily rising over the years. The accurate detection and classification of liver cancer are pivotal for timely treatment and improved patient outcomes. The most challenging tasks identified from the previous research studies are computational complexity, sensitive parameter setting, misdetection and misclassification. So, a deep learning-based optimization algorithm is proposed to detect and classify liver cancer. The image data are collected from the LiTS17 dataset, 3D-IRCADb dataset and Liver tumor CT dataset to preprocess the medical image data and the preprocessing provides consistency, quality and privacy. The Differential Convolutional neural Network (Differential CNN) model extracts the relevant features for improving the ability of a model to differentiate healthy and cancerous tissues. The features are classified into benign and malignant by using the classification model namely Kernel Extreme Learning Machine (KELM) model. The Differential Biogeography-Based Optimization Algorithm (DBBOA) algorithm fine-tunes the parameters to find near-optimal solutions. This tuning process is conducted during training the deep learning-based classification model. The experimental validation is conducted in terms of using significant performance evaluation measures and the comprehensive analysis provided a better classification accuracy of 98.72%, F1-score of 98.25%, specificity of 97.93%, sensitivity of 98.52%, AUCROC of 0.9872, precision of 98.89% and a computational time of 1.5 s for the proposed liver cancer detection and classification model. The comparative analysis showed that the proposed model achieved a superior outcome rather than other existing methods.

关键词： Liver cancer diagnosis Differential biogeography-based optimization algorithm Kernel extreme learning machine Differential convolutional neural network Optimal solutions Classification Feature extraction and high-dimensional pattern

来源：评论

学校读者我要写书评

暂无评论

MixUNet: A lightweight medical image segmentation network capturing multidimensional semantic information

引用

BIOMEDICAL signal processing AND CONTROL 2024年 96卷

作者： Chen, Yufeng Zhang, Xiaoqian He, Youdong Peng, Lifan Pu, Lei Sun, Feng Southwest Univ Sci & Technol Sch Informat Engn Mianyang 621010 Peoples R China Univ Elect Sci & Technol China Mianyang Cent Hosp Sch Med Mianyang 621010 Peoples R China Mianyang Cent Hosp NHC Key Lab Nucl Technol Med Transformat Mianyang 621010 Peoples R China

The efficient segmentation of medical image is of great significance for clinical diagnosis. Recently, TransUNet has achieved great success in medical image segmentation by effectively fusing Convolutional neural Networks (CNN) and Vision Transformer (ViT) to accomplish the extraction of local and global information. However, since TransUNet is designed as a stitching of CNN and ViT framework level, it has the following problems to be solved: 1) only local and relatively global spatial features of images are extracted;2) the direct introduction of ViT brings the disadvantages of not easy training and high computational overhead. Therefore, in this work, we propose Mixblock, a hybrid encoder that effectively fuses the superiority of CNN and ViT and is capable of extracting multidimensional high-level semantic information of images instead of being limited to local and global spatial features. Based on this, we design a UNet-like method MixUNet for medical image segmentation, which is a concise and efficient baseline network. Specifically, MixUNet is able to converge after less training without any pre-training, and its number of parameters and computation are only 3.17% and 4.99% of those of TransUNet. In addition, we creatively introduce frequency domain information on skip connection to eliminate the semantic ambiguity between the encoder and decoder, which provides a new perspective for medical image segmentation. Finally, we perform extensive experiments on three publicly available medical image datasets. Experimental results show that MixUNet has significant superiority in segmentation performance, model complexity, and robustness compared to state-of-the-art baseline methods.

关键词： Medical image segmentation UNet CNN ViT

来源：评论

学校读者我要写书评

暂无评论

REGIR: REFINED GEOMETRY FOR SINGLE-image IMPLICIT CLOTHED HUMAN RECONSTRUCTION 49

REGIR: REFINED GEOMETRY FOR SINGLE-IMAGE IMPLICIT CLOTHED HU...

引用

49th IEEE International Conference on Acoustics, Speech, and signal processing (ICASSP)

作者： Yao, Li Gao, Ao Wan, Yan Donghua Univ Sch Comp Sci & Technol Shanghai 201620 Peoples R China

ISBN: (纸本)9798350344868;9798350344851

Recently, implicit function-based approaches have advanced 3D human reconstruction from a single-view image. However, previous methods suffer from issues such as noisy artifacts, loss of geometric details, and broken limbs under the scenarios of challenging poses. To address these problems, a novel end-to-end deep neural network named ReGIR is proposed, which is a multi-level architecture combining the parametric model with implicit function. The architecture consists of a coarse level and a fine level, and for each level, normal maps and the signed distance function (SDF) are introduced to encode query points. Furthermore, the network is trained in a coarse-to-fine manner to enable robust human body reconstruction with geometric details. Our extensive qualitative and quantitative experiments demonstrate that ReGIR achieves competitive reconstruction results.

关键词： Single-view human reconstruction Parametric model Implicit function SDF

来源：评论

学校读者我要写书评

暂无评论

Alzheimer's Dementia Detection: An Optimized Approach using ITD of EEG signals 32

Alzheimer's Dementia Detection: An Optimized Approach using ...

引用

32nd European signal processing Conference (EUSIPCO)

作者： Sen, Sena Yagmur Akan, Aydin Cura, Ozlem Karabiber Izmir Univ Econ Dept Elect & Elect Engn Izmir Turkiye Izmir Katip Celebi Univ Dept Biomed Engn Izmir Turkiye

ISBN: (纸本)9789464593617;9798331519773

This paper presents a novel early-stage Alzheimer's dementia (AD) disease detection based on convolutional neural networks (CNNs). As it is widely used in detection and classification of AD disease, a time-frequency (TF) method has been proposed for AD detection. It has been described to address the problem of detecting early-stage AD by combining TF and CNN methods. The method is developed by utilizing the well-known structural similarity index measure (SSIM) to obtain discriminative features in each TF image. Experimental results demonstrate that the proposed method outperforms the early-stage AD detection using advanced signal decomposition algorithm that is intrinsic time-scale decomposition (ITD), and it achieves a notable improvement in terms of the detection success rates compared to AD detection from TF images of raw EEG signals.

关键词： Alzheimer's dementia (AD) Electroencephalography (EEG) Intrinsic Time-Scale Decomposition (ITD) Short-Time Fourier Transform (STFT) Convolutional neural Network (CNN)

来源：评论

学校读者我要写书评

暂无评论

A Fault Diagnosis Method of Rotor System Based on Parallel Convolutional neural Network Architecture with Attention Mechanism

引用

JOURNAL OF signal processing SYSTEMS FOR signal image AND VIDEO TECHNOLOGY 2023年第8期95卷 965-977页

作者： Zhao, Zhiqian Jiao, Yinghou Zhang, Xiang Harbin Inst Technol Sch Mechatron Engn Harbin 150000 Heilongjiang Peoples R China Harbin Inst Technol Lab Vibrat & Noise Control Harbin 150000 Heilongjiang Peoples R China

In practical engineering applications, the working load of the rotor system is changing constantly, and the noise pollution of its working environment is serious, which leads to the performance degradation of traditional fault diagnosis methods. To solve the above problems, we present a novel rotor system fault diagnosis model based on parallel convolutional neural network architecture with attention mechanism (AMPCNN). The model uses convolution kernels of different sizes in parallel channels to process raw data, and based on late feature fusion, a more comprehensive feature map is obtained. Furthermore, the information sharing between the two channels is realized through the attention mechanism so that the effective features of one channel can be reflected in another channel. The performance of the model under variable working conditions is verified by the Machinery Fault Database (MAFAULDA), and the average accuracy is 99.58%. By dividing Gaussian white noise from -9 dB to 2 dB into 11 intervals and adding it to the public data of Wuhan University, the noise resistance performance is verified, and the proposed method can obtain 100% diagnosis accuracy even in the high noise condition. The above experiments show that in terms of load adaptability and noise immunity, the method has higher accuracy than traditional deep learning classification methods.

关键词： Rotor system Fault diagnosis Feature fusion Convolutional neural network Attention mechanism

来源：评论

学校读者我要写书评

暂无评论

image enhancement with intensity transformation on embedding space

引用

CAAI Transactions on Intelligence Technology 2024年第1期9卷 101-115页

作者： Hanul Kim Yeji Jeon Yeong Jun Koh Department of Applied Artificial Intelligence Seoul National University of Science and TechnologySeoulSouth Korea Department of Computer Science and Engineering Chungnam National UniversityDaejeonSouth Korea

In recent times,an image enhancement approach,which learns the global transformation function using deep neural networks,has gained ***,many existing methods based on this approach have a limitation:their transformation functions are too simple to imitate complex colour transformations between low-quality images and manually retouched high-quality *** order to address this limitation,a simple yet effective approach for image enhancement is *** proposed algorithm based on the channel-wise intensity transformation is ***,this transformation is applied to the learnt embedding space instead of specific colour spaces and then return enhanced features to *** this end,the authors define the continuous intensity transformation(CIT)to describe the mapping between input and output intensities on the embedding ***,the enhancement network is developed,which produces multi-scale feature maps from input images,derives the set of transformation functions,and performs the CIT to obtain enhanced *** experiments on the MIT-Adobe 5K dataset demonstrate that the authors’approach improves the performance of conventional intensity transforms on colour space ***,the authors achieved a 3.8%improvement in peak signal-to-noise ratio,a 1.8%improvement in structual similarity index measure,and a 27.5%improvement in learned perceptual image patch ***,the authors’algorithm outperforms state-of-the-art alternatives on three image enhancement datasets:MIT-Adobe 5K,Low-Light,and Google HDRþ.

关键词： computer vision deep learning image enhancement image processing

来源：评论

学校读者我要写书评

暂无评论

Riemannian Generalized Gaussian Distributions on the Space of SPD Matrices for image Classification

引用

IEEE ACCESS 2024年 12卷 26096-26109页

作者： Abbad, Zakariae El Maliani, Ahmed Drissi El Hassouni, Mohammed Abbassi, Mohamed Tahar Kadaoui Bombrun, Lionel Berthoumieu, Yannick Mohammed V Univ Rabat ENSIAS Rabat 10000 Morocco Mohammed V Univ Rabat Fac Sci LRIT Rabat IT Ctr Rabat 10000 Morocco Mohammed V Univ Rabat FLSH Rabat 10000 Morocco Sidi Mohamed Ben Abdellah Univ Fac Sci Dhar El Mahraz Lab Math Sci & Applicat Fes 30000 Morocco Univ Bordeaux CNRS Bordeaux INP IMSUMR 5218 F-33400 Talence France Bordeaux Sci Agro F-33175 Gradignan France

The space of symmetric positive definite (SPD) matrices, denoted as $P_{m}$ , plays a crucial role in various domains, including computer vision, medical imaging, and signal processing. Its significance lies in its capacity to represent the underlying structure in nonlinear data using its Riemannian geometry. Nevertheless, a notable gap exists in the absence of statistical distributions capable of characterizing the statistical properties of data within this space. This paper proposes a new Riemannian Generalized Gaussian distribution (RGGD) on that space. The major contributions of this paper are, first of all, providing the exact expression of the probability density function (PDF) of the RGGD model, as well as an exact expression of the normalizing factor. Furthermore, an estimation of parameters is given using the maximum likelihood of this distribution. The second contribution involves exploiting the second-order statistics of feature maps derived from the first layers of deep convolutional neural networks (DCNNs) through the RGGD stochastic model in an image classification framework. Experiments were carried out on four well-known datasets, and the results demonstrate the efficiency and competitiveness of the proposed model.

关键词： Symmetric positive definite matrices generalized Gaussian distribution texture Riemannian geometry Rao's distance Riemannian metric

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：