检索结果-内蒙古大学图书馆

Affine registration of thermal images of plantar feet using convolutional neural networks

BIOMEDICAL signal processing AND CONTROL 2024年第PartB期95卷

作者： Aferhane, Asma Bouallal, Doha Douzi, Hassan Harba, Rachid Vilcahuaman, Luis Arbanil, Hugo Ibn Zohr Univ IRF SIC Lab Agadir Morocco Orleans Univ PRISME Lab Orleans France PUCP Univ Lima Peru Hosp Natl Mayo Lima Peru

The use of a thermal camera to detect abnormal plantar foot temperature changes can be an effective way to identify the early signs of diabetic foot ulceration. In this work, we performed the affine registration of the plantar foot thermal images using four models based on convolutional neural networks. The process include two parts: an affine registration model for estimating transformation parameters and a spatial transformer for getting the registered image. The performances of the four models were evaluated using the Dice similarity coefficient (DSC), Mean Square Error (MSE), and peak signal-to-noise ratio (PSNR). In the first step, methods were applied to register the left and right feet of the same subject, called ''contralateral registration'' and in the second step, the methods were evaluated on a pair of images of the same subject taken in two different times (T-0 and T-10) using a cold stress test protocol. Results showed that the used convolutional neural networks are robust in both types of registration (contralateral and multitemporal), and they are suitable for the targeted application, with the DSC of 95% for contralateral registration and a DSC of 92% for multitemporal registration. Furthermore, a transversal clinical study was perform on diabetic patients, that classified individuals into ischemic and non-ischemic groups. The objective was to analyze the coherence between the thermal results and medical data. The mean absolute point-to-point temperature difference |Delta T| between left and right foot is lower in non-ischemic patients than in those with ischemia, with p < 0.05.

关键词： Plantar foot thermal images Deep learning Thermography Medical images Affine registration Convolutional neural network

来源：评论

学校读者我要写书评

暂无评论

Removal of Speckle Noises from Ultrasound images Using Parallel Convolutional neural Network

引用

CIRCUITS SYSTEMS AND signal processing 2023年第8期42卷 5041-5064页

作者： Shen, Zhengjie Tang, Chen Xu, Min Lei, Zhenkun Tianjin Univ Sch Elect & Informat Engn Tianjin 300072 Peoples R China Dalian Univ Technol State Key Lab Struct Anal Ind Equipment Dalian 116024 Peoples R China

Speckle noises widely exist in ultrasound images. They seriously affect the quality of images and cause the doctor to make mistakes in diagnosis. In this paper, we propose a three-path parallel convolutional neural network called USNet to achieve speckle reduction for ultrasound images. We combine three different sub-networks to increase the width of the whole network instead of the depth. The ideas of dilated convolution and shortcut connection are added to increasing the learning ability of themodel. These make our proposed USNet can learn deeper information from the original ultrasound image. In experiments, we verify the effectiveness of the proposed three-path parallel structure and the dilated convolution by conducting ablation experiments. At the same time, we propose a different method to construct the training dataset. For the noisy training image, we artificially add speckle noise at three different s levels to enhance the generalization performance of the proposed method. For the noise-free true labels, we use two stages to obtain on the basis of original images, including Optimized BayesianNon-Local Means with block selectionmethod (OBNLM) and Second-order Oriented Partial-differential Equation (SOOPDE) method. Then we compare our proposed method with other four different methods, including Kuan method, Speckle Reducing Anisotropic Diffusion (SRAD) method, the OBNLM method, and Residual Learning Network (RLNet). We qualitatively and quantitatively evaluate these methods in terms of smoothness, texture information protection, and edge clarity. The results show that our proposed USNet model can batch and quickly achieve good speckle reduction as well as texture preservation for different types of ultrasound images without any parameter adjustment. The USNet model has the advantages of good adaptability, robustness, and generalization. It is of great significance for improving the diagnostic efficiency of clinical medicine.

关键词： Ultrasound Speckle noise image processing Convolution neural network

来源：评论

学校读者我要写书评

暂无评论

LULC Change Detection Using Combined Machine and Deep Learning Classifiers 7

LULC Change Detection Using Combined Machine and Deep Learni...

引用

IEEE 7th International Conference on Advanced Technologies, signal and image processing (ATSIP)

作者： Tahraoui, Ahmed Kheddam, Radja USTHB Image Proc & Radiat Lab Fac Elect Engn Algiers Algeria

ISBN: (纸本)9798350351491;9798350351484

In this paper we shall describe a method for land use and land cover (LULC) change detection in multi-date multispectral images using a well known post-classification comparison (PCC) principle. The novelty of the proposed method is the use of a random forest (RF) classifier and a deep neural network (DNN) classifier to obtain the multi-temporal land use maps required to perform a PCC process. Recall that RF is a powerful and versatile supervised machine learning algorithm, and DNN is considered as the most basic and effective deep learning architecture. A mono-date multi-classifiers Bayesian decision fusion is executed to enhance the input LULC mapping of the PCC scheme. Co-registered bi-temporal multispectral images taken over the northeastern region of Algiers, Algeria, between 1997 and 2001 by the American LANDSAT-TM satellite are used to verify the effectiveness of the proposed method which is also compared to the more elaborated convolutional neural network (CNN) method. While recently developed deep learning change detection methods (DLCDM) necessitate large labeled training data and place high demands on computational power in hardware and software, the PCC scheme we proposed is simple to implement, does not consume a lot of computation time and produces results that are consistent with the ground truth of the study zone.

关键词： Change detection multi-temporal images Random forest (RF) deep neural network (DNN)

来源：评论

学校读者我要写书评

暂无评论

Detection and Classification of Neurodegenerative Diseases by Automatic Speech Analysis 7

Detection and Classification of Neurodegenerative Diseases b...

引用

IEEE 7th International Conference on Advanced Technologies, signal and image processing (ATSIP)

作者： Kehili, Ahlem Bouafif, Lamia Cherif, Adnen Univ Tunis Manar ATSSEE Lab FST Tunis 2092 Tunisia

ISBN: (纸本)9798350351491;9798350351484

This study presents a novel non-invasive method for detecting and classifying neuro-degenerative diseases such as Parkinson's (PD) and Alzheimer's (AD) through automatic speech analysis and artificial intelligence. The analysis of the voice recordings was carried out using different parametric extraction methods based on the MFCC and prosodic coefficients (VOT, Jitter, Shimmer, HNR, ... ) followed by a classification step based on CNN and FC-DNN neural network. These methods made it possible to extract relevant speech parameters and use them for training and classification. The results obtained showed vocal disturbances in mild and preclinical stages of PD and AD such as articulation, prosody and rhythmic abilities. Developed machine learning algorithms were able to detect subjects with PD with 98% accuracy from rapid syllable repetitions and 96% accuracy for subjects with AD from voice parameters.

关键词： Disease detection Parkinson & Alzheimer speech processing Artificial intelligence classification

来源：评论

学校读者我要写书评

暂无评论

An efficient deep convolutional neural network with features fusion for radar signal recognition

引用

MULTIMEDIA TOOLS AND APPLICATIONS 2023年第2期82卷 2871-2885页

作者： Si, Weijian Wan, Chenxia Deng, Zhian Harbin Engn Univ Coll Informat & Commun Engn Harbin 150001 Peoples R China Harbin Engn Univ Key Lab Adv Marine Commun & Informat Technol Minist Ind & Informat Technol Harbin 150001 Peoples R China

This paper proposes an efficient deep convolutional neural network with features fusion for recognizing radar signal, which mainly includes data pre-processing, features extraction, multi-features fusion, and classification. Radar signals are first transformed into time-frequency images by using choi-williams distribution and smooth pseudo-wigner-ville distribution, and the image pre-processing methods are used to resize and normalize the time-frequency images. Then, two constructed deep convolutional neural network models are aimed to extract more effective features. Furthermore, a multi-features fusion model is proposed to integrate features extracted from two deep convolutional neural network models, which makes full use of the relationship among different features and further improves the recognition performance. Experimental results shown that the average recognition accuracy of the proposed method is up to 84.38% when the signal to noise ratio is at -12 dB, and even reach to 94.31% at -10 dB, which achieved the superior recognition performance than others, especially at the lower signal to noise ratio. Moreover, the recognition performance of various radar signals can be largely improved, especially for 2FSK, 4FSK and SFM. This work provides a sound experimental foundation for further improving radar signal recognition in modern electronic warfare systems.

关键词： Radar signal recognition Convolutional neural network Features fusion signal to noise ratio Recognition accuracy

来源：评论

学校读者我要写书评

暂无评论

Binary stochastic Flip Optimization for Training Binary neural Networks

Binary Stochastic Flip Optimization for Training Binary Neur...

引用

2025 IEEE International Conference on Acoustics, Speech, and signal processing, ICASSP 2025

作者： Shibuya, Tatsukichi Inoue, Nakamasa Kawakami, Rei Sato, Ikuro Department of Computer Science Institute of Science Tokyo Tokyo Japan Department of Systems and Control Engineering Institute of Science Tokyo Tokyo Japan Denso IT Laboratory Tokyo Japan

ISBN: (纸本)9798350368741

For deploying deep neural networks on edge devices with limited resources, binary neural networks (BNNs) have attracted significant attention, due to their computational and memory efficiency. However, once a neural network is binarized, finetuning it on edge devices becomes challenging because most conventional training algorithms for BNNs are designed for use on centralized servers and require storing real-valued parameters during training. To address this limitation, this paper introduces binary stochastic flip optimization (BinSFO), a novel training algorithm for BNNs. BinSFO employs a parameter update rule based on Boolean operations, eliminating the need to store real-valued parameters and thereby reducing memory requirements and computational overhead. In experiments, we demonstrated the effectiveness and memory efficiency of BinSFO in finetuning scenarios on six image classification datasets. BinSFO performed comparably to conventional training algorithms with a 70.7% smaller memory requirement. Code is released at https://***/TatsukichiShibuya/ICASSP2025_BinSFO. © 2025 IEEE.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Visual Cardiac signal Classifiers: A Deep Learning Classification Approach for Heart signal Estimation From Video

引用

IEEE ACCESS 2024年 12卷 144377-144394页

作者： Moustafa, Mohamed Farooq, Muhammad Ali Elrasad, Amr Lemley, Joseph Corcoran, Peter Univ Galway Dept Elect & Elect Engn Imaging Lab C3I Galway H91 TK33 Ireland FotoNation Sensing Team Galway H91 V0TX Ireland

Heart rate is a crucial metric in health monitoring. Traditional computer vision solutions estimate cardiac signals by detecting physical manifestations of heartbeats, such as facial discoloration caused by blood oxygenation changes, from subject videos using regression methods. As continuous signals are more complex and expensive to de-noise, this study introduces an alternative approach, employing end-to-end classification models to remotely derive a discrete representation of cardiac signals from face videos. These visual cardiac signal classifiers are trained on discretized cardiac signals, a novel pre-processing method with limited precedent in health monitoring literature. Consequently, various methods to convert continuous cardiac signals into binary form are presented, and their impact on training is evaluated. An implementation of this approach, the temporal shift convolutional attention binary classifier, is presented using the regression-based convolutional attention network architecture. The classifier and a baseline regression model are trained and tested using publicly available and locally collected datasets designed for heart signal detection from face video. The model performance is then assessed based on the heart rate error from the extracted cardiac signals. Results show the proposed method outperforms the baseline on the UBFC-rPPG dataset, reducing cross-dataset root mean square error from 2.33 to 1.63 beats per minute. However, both models struggled to generalize to the PURE dataset, with root mean square errors of 12.40 and 16.29 beats per minute, respectively. Additionally, the proposed approach reduces the computational complexity of model output post-processing, enhancing its suitability for real-time applications and deployment on systems with restricted resources.

关键词： Convolutional neural networks Heart rate Computer vision Training Estimation Remote sensing Visualization Standards Monitoring Feature extraction Deep learning Biomedical signal processing Photoplethysmography deep learning image classification remote monitoring remote sensing biomedical signal processing heart rate photoplethysmography

来源：评论

学校读者我要写书评

暂无评论

EBSTracker: event-based space multi-object tracking with bidirectional self-attention and multi-stage data association

引用

signal image AND VIDEO processing 2025年第1期19卷 1-9页

作者： Zhou, Xiaoli Bei, Chao CASIC Space Engn Dev Co Ltd Beijing 100854 Peoples R China Second Res Acad CASIC Grad Sch Beijing 100854 Peoples R China

Event cameras are preferred for space object tracking due to their high temporal resolution and ability to capture dim light, fast-moving objects, and other challenging space objects. However, existing event trackers still use conventional tracking methods based on object textures, which may not be robust enough for challenging scenarios. To address this, we propose the Event-Based Space multi-object Tracker (EBSTracker), which integrates a bidirectional self-attention network and a multi-stage data association network. The bidirectional self-attention network enhances feature representation for tiny objects, while the multi-stage data association network uses the Noise Scale Adaptive (NSA) Kalman filter and Generalized Intersection over Union (GIoU) metric to predict trajectory positions, improving tracking robustness. Experiments on two large-scale datasets have demonstrated the effectiveness and robustness of EBSTracker, achieving state-of-the-art (SOTA) performance in challenging scenarios with tiny moving objects. This has advanced event-based space multi-object tracking technology.

关键词： Multi-object tracking (MOT) Convolutional neural networks Event-based MOT datasets. Event camera

来源：评论

学校读者我要写书评

暂无评论

A semi-supervised framework with generative adversarial network for pansharpening

引用

signal image AND VIDEO processing 2025年第6期19卷 1-12页

作者： Wang, Yu-Xuan Huang, Ting-Zhu Ran, Ran Wen, Rui Deng, Liang-Jian Univ Elect Sci & Technol China Chengdu 611731 Sichuan Peoples R China

The pansharpening method of fusing remote sensing satellite photography to obtain higher quality images is an increasingly hot research topic. However, the scarcity of the ground truth makes it difficult to conduct supervised learning with large amounts of data. In this paper, we propose a multi-scale network for semi-supervised learning (PSSGNet), in which we adopt generative adversarial network (GAN) for the extracted features of each layer to reduce the domain differences on reduced and full resolution data. Our framework is mainly divided into two parts. Firstly, we propose a multi-scale framework to extract informative features and restore more spectral information from images of different scales, thus realizing the fusion and complementary of features of multiple images. Secondly, we apply GAN to the output of each layer to eliminate the negative impact caused by inherent domain gap, with which we obtain relevant parameters to migrate to the training of real images and use a small amount of real data to lead the reduced resolution data to compensate for the missing information. Extensive experiments show that compared with other recent competitive methods, the proposed one has certain advantages in the quantitative metrics especially on full resolution data, and also achieves better visual results.

关键词： Pansharpening Convolutional neural network Semi-supervised learning GAN

来源：评论

学校读者我要写书评

暂无评论

Multi-scale Unet-based feature aggregation network for lightweight image deblurring

引用

signal image AND VIDEO processing 2025年第1期19卷 1-10页

作者： Yang, Yancheng Gai, Shaoyan Da, Feipeng Southeast Univ Sch Automat Nanjing 210096 Jiangsu Peoples R China Minist Educ Key Lab Measurement & Control Complex Syst Engn Nanjing 210096 Jiangsu Peoples R China

The single image deblurring task has made remarkable progress, with convolutional neural networks exhibiting extraordinary performance. However, existing methods maintain high-quality reconstruction through an excessive number of parameters and extremely deep network structures, which results in increased requirements for computational resources and memory storage, making it challenging to deploy on resource-constrained devices. Numerous experiments indicate that current models still possess redundant parameters. To address these issues, we introduce a multi-scale Unet-based feature aggregation network (MUANet). This network architecture is based on a single-stage Unet, which significantly simplifies the network's complexity. A lightweight Unet-based attention block is designed, based on a progressive feature extraction module to enhance feature extraction from multi-scale attention modules. Given the extraordinary performance of the self-attention mechanism, we propose a self-attention mechanism based on fourier transform and a depthwise convolutional feed-forward network to enhance the network's feature extraction capability. This module contains extractors with different receptive fields for feature extraction at different spatial scales and capturing contextual information. Through the aggregation of multi-scale features from different attention mechanisms, our method learns a set of rich features that retain contextual information from multiple scales and high-resolution spatial details. Extensive experiments show that the proposed MUANet achieves competitive results in lightweight deblurring qualitative and quantitative evaluations.

关键词： Lightweight image deblurring Multi-scale features Attention mechanisms High-resolution spatial details

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：