检索结果-南通市图书馆

Speech Enhancement via Mask-Mapping Based Residual Dense Network

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2023年第1期74卷 1259-1277页

作者： Lin Zhou Xijin Chen Chaoyan Wu Qiuyue Zhong Xu Cheng Yibin Tang School of Information Science and Engineering Southeast UniversityNanjing210096China center for machine vision and signal analysis University of OuluOuluFI-90014Finland College of IOT Engineering Hohai UniversityChangzhou213022China

Masking-based and spectrum mapping-based methods are the two main algorithms of speech enhancement with deep neural network(DNN).But the mapping-based methods only utilizes the phase of noisy speech,which limits the upper bound of speech enhancement performance.Maskingbased methods need to accurately estimate the masking which is still the key problem.Combining the advantages of above two types of methods,this paper proposes the speech enhancement algorithm MM-RDN(maskingmapping residual dense network)based on masking-mapping(MM)and residual dense network(RDN).Using the logarithmic power spectrogram(LPS)of consecutive frames,MM estimates the ideal ratio masking(IRM)matrix of consecutive frames.RDN can make full use of feature maps of all layers.Meanwhile,using the global residual learning to combine the shallow features and deep features,RDN obtains the global dense features from the LPS,thereby improves estimated accuracy of the IRM matrix.Simulations show that the proposed method achieves attractive speech enhancement performance in various acoustic environments.Specifically,in the untrained acoustic test with limited priors,e.g.,unmatched signal-to-noise ratio(SNR)and unmatched noise category,MM-RDN can still outperform the existing convolutional recurrent network(CRN)method in themeasures of perceptual evaluation of speech quality(PESQ)and other evaluation indexes.It indicates that the proposed algorithm is more generalized in untrained conditions.

关键词： Mask-mapping-based method residual dense block speech enhancement

Distinguishing Between Natural and GAN-Generated Face Images by Combining Global and Local Features

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2022年第1期31卷 59-67页

作者： CHEN Beijing TAN Weijin WANG Yiting ZHAO Guoying Engineering Research center of Digital Forensics Ministry of Education Nanjing University of Information Science and Technology School of Computer Nanjing University of Information Science and Technology Jiangsu Collaborative Innovation center of Atmospheric Environment and Equipment Technology (CICAEET) Nanjing University of Information Science and Technology Warwick Manufacturing Group University of Warwick center for machine vision and signal analysis University of Oulu

With the development of face image synthesis and generation technology based on generative adversarial networks(GANs), it has become a research hotspot to determine whether a given face image is natural or generated. However, the generalization capability of the existing algorithms is still to be improved. Therefore,this paper proposes a general algorithm. To do so, firstly,the learning on important local areas, containing many face key-points, is strengthened by combining the global and local features. Secondly, metric learning based on the Arc Face loss is applied to extract common and discriminative features. Finally, the extracted features are fed into the classification module to detect GAN-generated faces.The experiments are conducted on two publicly available natural datasets(Celeb A and FFHQ) and seven GANgenerated datasets. Experimental results demonstrate that the proposed algorithm achieves a better generalization performance with an average detection accuracy over 0.99 than the state-of-the-art algorithms. Moreover, the proposed algorithm is robust against additional attacks, such as Gaussian blur, and Gaussian noise addition.

关键词： Generated image Global feature Local features Generative adversarial network Metric learning

Microphone Array Speech Separation Algorithm Based on TC-ResNet

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2021年第11期69卷 2705-2716页

作者： Lin Zhou Yue Xu Tianyi Wang Kun Feng Jingang Shi School of Information Science and Engineering Southeast UniversityNanjing210096China center for machine vision and signal analysis University of OuluOuluFI-90014Finland

Traditional separation methods have limited ability to handle the speech separation problem in high reverberant and low signal-to-noise ratio(SNR)environments,and thus achieve unsatisfactory results.In this study,a convolutional neural network with temporal convolution and residual network(TC-ResNet)is proposed to realize speech separation in a complex acoustic environment.A simplified steered-response power phase transform,denoted as GSRP-PHAT,is employed to reduce the computational cost.The extracted features are reshaped to a special tensor as the system inputs and implements temporal convolution,which not only enlarges the receptive field of the convolution layer but also significantly reduces the network computational cost.Residual blocks are used to combine multiresolution features and accelerate the training procedure.A modified ideal ratio mask is applied as the training target.Simulation results demonstrate that the proposed microphone array speech separation algorithm based on TC-ResNet achieves a better performance in terms of distortion ratio,source-to-interference ratio,and short-time objective intelligibility in low SNR and high reverberant environments,particularly in untrained situations.This indicates that the proposed method has generalization to untrained conditions.

关键词： Residual networks temporal convolution neural networks speech separation

Saliency-Based Segmentation of Optic Disc in Retinal Images

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2019年第1期28卷 71-75页

作者： ZOU Beiji LIU Qing YUE Kejuan CHEN Zailiang CHEN Jie ZHAO Guoying School of Information Science and Engineering Central South University Hunan First Normal University School of Information Science and Engineering center for machine vision and signal analysis University of Oulu

Accurate segmentation of Optic disc(OD)is significant for the automation of retinal analysis and retinal diseases screening. This paper proposes a novel optic disc segmentation method based on the saliency. It includes two stages: optic disc location and saliency-based segmentation. In the location stage, the OD is detected using a matched template and the density of the vessels.In the segmentation stage, we treat the OD as the salient object and formulate it as a saliency detection problem.To measure the saliency of a region, the boundary prior and the connectivity prior are exploited. Geodesic distance to the window boundary is computed to measure the cost the region spends to reach the window boundary.After a threshold and ellipse fitting, we obtain the OD.Experimental results on two public databases for OD segmentation show that the proposed method achieves thestate-of-the-art performance.

关键词： Retinal fundus Optic disc segmentation Boundary prior Region connection Saliency measure

NLWSNet:a weakly supervised network for visual sentiment analysis in mislabeled web images

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Information Technology & Electronic Engineering 2020年第9期21卷 1321-1333页

作者： Luo-yang XUE Qi-rong MAO Xiao-hua HUANG Jie CHEN Computer Science and Communication Engineering.Jiangsu University Zhenjiang 212013China Jiangsu Key Laboratory of Security Technology for Industrial Cyberspace Zhenjiang 212013China School of Computer Engineering Narjing Institute of TechnologyNanjing 211167China center for machine vision and signal analysis University of OulsOulu 8000Finland

Large-scale datasets are driving the rapid developments of deep convolutional neural networks for visual sentiment analysis.However,the annotation of large-scale datasets is expensive and time consuming.Instead,it iseasy to obtain weakly labeled web images from the Internet.However,noisy labels st.ill lead to seriously degraded performance when we use images directly from the web for training networks.To address this drawback,we propose an end-to-end weakly supervised learning network,which is robust to mislabeled web images.Specifically,the proposed attention module automatically eliminates the distraction of those samples with incorrect labels bv reducing their attention scores in the training process.On the other hand,the special-class activation map module is designed to stimulate the network by focusing on the significant regions from the samples with correct labels in a weakly supervised learning approach.Besides the process of feature learning,applying regularization to the classifier is considered to minimize the distance of those samples within the same class and maximize the distance between different class centroids.Quantitative and qualitative evaluations on well-and mislabeled web image datasets demonstrate that the proposed algorithm outperforms the related methods.

关键词： Visual sentiment analysis Weakly supervised learning Mislabeled samples Significant sentiment regions

维普期刊数据库

Separation and Visualization of Arteries and Heart in3D Computed Tomography Angiography Images

在线全文

学校读者我要写书评

暂无评论

Journal of Shanghai Jiaotong university(Science) 2017年第1期22卷 1-9页

作者：兰守忍崔超毅刘鑫 JOHNSON Shane 苏佳良陈本智王利生 Department of Automation Shanghai Jiao Tong University Department of Vascular Surgery Shanghai Ninth People’s Hospital Shanghai Jiao Tong University School of Medicine center for machine vision and signal analysis University of Oulu University of Michigan-Shanghai Jiao Tong University Joint Institute

The visualization of arteries and heart usually plays a crucial role in the clinical diagnosis, but researchers face the problems of region selection and mutual occlusion in clinical visualization. Therefore, the arteries and the heart cannot be easily visualized by current visualization methods. To solve the problems, we propose a new framework for arteries and cardiac visualization by combining a priori knowledge and the set operations.Firstly, a suitable region can be easily determined in the transfer function space with a priori knowledge and the visual feedback results. Secondly, the arteries and the heart can be directly extracted by the marked seed point.Finally, the arteries and the heart are separated for solving mutual occlusion through the set operations. This framework can easily solve the mutual occlusion problem in clinical visualization and greatly improve the region selection method in the transfer function space. Its effectiveness has been demonstrated on the basis of many experimental results.

关键词：吸藏移动多地图集交互抽取集合操作显示的体积 TP 391.9 A

维普期刊数据库

Transcriptome profiling and RXR gene family identification reveals the molecular mechanism of rapid aging after spawning of cuttlefish Sepiella japonica

在线全文

学校读者我要写书评

暂无评论

Journal of Oceanology and Limnology 2024年第3期42卷 865-880页

作者： Zhenyu DONG Jiemei ZHAO Feng GUO Shuangrui LIN Huai YANG Yingying YE Changfei CHI Hongfei LI Baoying GUO College of Marine Science and Technology Zhejiang Ocean UniversityZhoushan 316022China National Engineering Research center for Marine Aquaculture Zhejiang Ocean UniversityZhoushan 316022China Donghai Laboratory Zhoushan 316021China

Sepiella japonica is a worldwide marine cuttlefish species of high economic value.S.japonica routinely modifying behaviors in reproductive life,such as rapid aging until death after spawning,has been recognized in artificial breeding.However,reproductive behavior at the level of genes is rarely reported,thus,the research on the genetic basis of behavior,reproduction,and artificial breeding was limited.We applied RNA-seq in different stages of reproduction to investigate the reason of rapid aging after spawning,pre-maturity,pre-spawning after maturity,and post-spawning.The retinoid X receptor(RXR)gene family in S.japonica was identified,and 1343–1452 differentially expressed genes(DEGs)in all 3 stages of reproductive life were identified from pairwise m RNA comparisons.Furthermore,through the GO term and KEGG analysis,S.japonica could handle neuronal development and network formation before maturity and have a functional degradation of neural communication,signal transduction,vision,and gene expression after spawning.Eight Sj RXRαs have been identified and they played different roles in growth development or reproduction.Therefore,the regulation of several channels and receptors is the intrinsic molecular mechanism of rapid aging after spawning in S.japonica.This study revealed the survival strategy and provided fundamental data on the level of genes for understanding the reproductive behavior and the reproduction of S.japonica.

关键词： Sepiella japonica RAN-seq retinoid X receptor(RXR)gene family rapid aging intrinsic molecular mechanism

维普期刊数据库

An Interfacial Gel Electrode Patch with Tunable Hydrogen Bond Network for Electromyographic Sensing and Discrimination

在线全文

学校读者我要写书评

暂无评论

CCS Chemistry 2024年第2期6卷 450-464页

作者： Jianliang Li Zhihao Liu Yongtao Tang Jiabao Xian Chaofan He Hao Wu Mingjie Liu Fengyu Li College of Chemistry and Materials Science Guangdong Provincial Key Laboratory of Functional Supramolecular Coordination Materials and ApplicationsSu Bingtian Center for Speed Research and TrainingJinan UniversityGuangzhou 510632 School of Chemistry Beihang UniversityBeijing 100191

The sensitivity and fidelity of surface electromyography(sEMG)signal monitoring is critical for muscle status and fatigue assessment,prosthetic control,and gesture recognition.However,the incompatible skin-electrode interface and complex electrophysiological environment restrict the sensitive acquisition and accurate analysis of sEMG signals.Focused on the impedance of the skin-electrode interface issue,we developed an interfacial gel electrode patch with a tunable hydrogen bond network to simultaneously achieve a conformal interface,suitable adhesion,and high conductivity for sEMG signal monitoring.By exploiting hydroxyethylidene diphosphonic acid(HEDP)and 2-hydroxyphosphono-acetic acid(HPAA)as hydrogen bonding regulators were introduced into the polyvinyl alcohol(PVA)-based hydrogel network to regulate the hydrogen bond cross-linking network.As a result,the balance of elastic modulus,adhesion,and electrical conductivity of PVA-HEDP-HPAA(PHH)hydrogel are achieved.The reliable electrodeskin interface is manipulated to achieve conformal contact by matching the elastic modulus,reducing the gap of electrode-skin interface by adhesion,and promoting ion and electron conduction by electrical conductivity.The PHH electrode patches exhibit a lower interfacial impedance(12.56 kΩ)and a signal-to-noise ratio of 38.09±1.28 dB for accurate analysis of muscle strength and evaluation of the fatigue state.With the assistance of the artificial neural network algorithm,seven gestures can be recognized with 100%prediction accuracy.The interfacial gel electrode patch contributes a bio-matching electrophysiological platform for prosthetic control,human–machine interaction,and clinical or athletic auxiliary monitoring.

关键词： hydrogel electrode hydrogen bond network surface electromyography muscle fatigue artificial neural network gesture recognition

Deep Learning and Its Application in Diabetic Retinopathy Screening

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2020年第6期29卷 992-1000页

作者： ZOU Beiji SHAN Xi ZHU Chengzhang DAI Yulan YUE Kejuan CHEN Yuanqiong XIAO Yalong HUANG Jiaer School of Computer Science and Engineering Central South University Hunan Engineering Research center of machine vision and Intelligent Medicine Central South University The College of Literature and Journalism Central South University

Deep learning(DL),especially Convolutional neural networks(CNN),has gained wide popularity in various image processing tasks.With the significant achievements obtained in DL,it has provided many successful solutions for real-world applications as well as in medical domain.Automated retinal images analysis has been widely applied to screening Diabetic retinopathy(DR),which can greatly help preventing the occurrence of complete blindness when used in the early screening.In this paper,we mainly focus on DL,and we will give an overview of the deep learning-based methods for DR screening.Finally,we will discuss the main issues encountered in the DR screening systems.

关键词： Deep learning Convolutional neural networks Diabetic retinopathy screening Lesion detection