Resolvers are normally employed for rotor positioning in motors for electric vehicles, but resolvers are expensive and vulnerable to vibrations. Hall sensors have the advantages of low cost and high reliability, but t...Resolvers are normally employed for rotor positioning in motors for electric vehicles, but resolvers are expensive and vulnerable to vibrations. Hall sensors have the advantages of low cost and high reliability, but the positioning accuracy is low. Motors with Hall sensors are typically controlled by six-step commutation algorithm, which brings high torque ripple. This paper studies the high-performance driving and braking control of the in-wheel permanent magnetic synchronous motor （PMSM） based on low-resolution Hall sensors. Field oriented control （FOC） based on Hall-effect sensors is developed to reduce the torque ripple. The positioning accuracy of the Hall sensors is improved by interpolation between two consecutive Hall signals using the estimated motor speed. The position error from the misalignment of the Hall sensors is compensated by the precise calibration of Hall transition timing. The braking control algorithms based on six-step commutation and FOC are studied. Two variants of the six-step commutation braking control, namely, half-bridge commutation and full-bridge commutation, are discussed and compared, which shows that the full-bridge commutation could better explore the potential of the back electro-motive forces （EMF）, thus can deliver higher efficiency and smaller current ripple. The FOC braking is analyzed with the phasor diagrams. At a given motor speed, the motor turns from the regenerative braking mode into the plug braking mode if the braking torque exceeds a certain limit, which is proportional to the motor speed. Tests in the dynamometer show that a smooth control could be realized by FOC driving control and the highest efficiency and the smallest current ripple could be achieved by FOC braking control, compared to six-step commutation braking control. Therefore, FOC braking is selected as the braking control algorithm for electric vehicles. The proposed research ensures a good motor control performance while maintaining low cost and high reliability.展开更多
A comparison study is performed to contrast the improvements in the tropical Pacific oceanic state of a low-resolution model respectively via data assimilation and by an increase in horizontal resolution. A low resolu...A comparison study is performed to contrast the improvements in the tropical Pacific oceanic state of a low-resolution model respectively via data assimilation and by an increase in horizontal resolution. A low resolution model (LR) (1°lat by 2°lon) and a high-resolution model (HR) (0.5°lat by 0.5°lon) are employed for the comparison. The authors perform 20-yr numerical experiments and analyze the annual mean fields of temperature and salinity. The results indicate that the low-resolution model with data assimilation behaves better than the high-resolution model in the estimation of ocean large-scale features. From 1990 to 2000, the average of HR's RMSE (root-mean-square error) relative to independent Tropical Atmosphere Ocean project (TAO) mooring data at randomly selected points is 0.97℃ compared to a RMSE of 0.56℃ for LR with temperature assimilation. Moreover, the LR with data assimilation is more frugal in computation. Although there is room to improve the high-resolution model, the low-resolution model with data assimilation may be an advisable choice in achieving a more realistic large-scale state of the ocean at the limited level of information provided by the current observational system.展开更多
In a millimeter-wave(mmWave)Massive multiple-input multiple-output(MIMO)systems,full-digital beamforming(i.e.,connecting each antenna with a specific radio-frequency(RF)chain)becomes inefficient due to the hardware co...In a millimeter-wave(mmWave)Massive multiple-input multiple-output(MIMO)systems,full-digital beamforming(i.e.,connecting each antenna with a specific radio-frequency(RF)chain)becomes inefficient due to the hardware cost and power consumption.Therefore,hybrid analog and digital transceiver where the number of RF chains are much smaller than that of the antennas has drawn great research interest.In this work,we investigate the use of low-resolution analog-to-digital converters(ADCs)in the uplink of multi-user hybrid and full-digital mmWave Massive MIMO systems.To be specific,we compare the performance of full-digital minimum mean square error(MMSE)and hybrid MMSE beamforming in both sum rates and energy efficiency.Accurate approximations of sum rates and energy efficiency are provided for both schemes,which captures the dominant factors.The analytical results show that full-digital beamforming outperforms hybrid beamforming in terms of sum rates and requires only a small portion(γ)of antennas used by hybrid beamforming to achieve the same sum rates.We given sufficient condition for full-digital beamforming to outperform hybrid beamforming in terms of energy efficiency.Moreover,an algorithm is proposed to search for the optimal ADC resolution bits.Numerical results demonstrate the correctness of the analysis.展开更多
In this paper, we focus on low-resolution human detection and propose a partial least squares-canonical correlation analysis （PLS-CCA） for outdoor video surveillance. The analysis relies on heterogeneous features fu...In this paper, we focus on low-resolution human detection and propose a partial least squares-canonical correlation analysis （PLS-CCA） for outdoor video surveillance. The analysis relies on heterogeneous features fusion-based human detection method. The proposed method can not only explore the relation between two individual heterogeneous features as much as possible, but also can robustly describe the visual appearance of humans with complementary information. Compared with some other methods, the experimental results show that the proposed method is effective and has a high accuracy, precision, recall rate and area under curve （AUC） value at the same time, and offers a discriminative and stable recognition performance.展开更多
IRAS low-resolution spectra are presented for 36 HAEBE stars. It is found that silicate dust in amorphous or glassy form is common material in the circumstellar disks or/and shells of HAEBE stars. It is also found tha...IRAS low-resolution spectra are presented for 36 HAEBE stars. It is found that silicate dust in amorphous or glassy form is common material in the circumstellar disks or/and shells of HAEBE stars. It is also found that the PAH feature is often appeared as well.展开更多
Nonparametric time-of-arrival（TOA） estimators for impulse radio ultra-wideband（IR-UWB） signals are proposed. Nonparametric detection is obviously useful in situations where detailed information about the statistic...Nonparametric time-of-arrival（TOA） estimators for impulse radio ultra-wideband（IR-UWB） signals are proposed. Nonparametric detection is obviously useful in situations where detailed information about the statistics of the noise is unavailable or not accurate. Such TOA estimators are obtained based on conditional statistical tests with only a symmetry distribution assumption on the noise probability density function. The nonparametric estimators are attractive choices for low-resolution IR-UWB digital receivers which can be implemented by fast comparators or high sampling rate low resolution analog-to-digital converters（ADCs）,in place of high sampling rate high resolution ADCs which may not be available in practice. Simulation results demonstrate that nonparametric TOA estimators provide more effective and robust performance than typical energy detection（ED） based estimators.展开更多
The uplink achievable rate of massive multiple-input-multiple-output(MIMO) systems, where the low-resolution analog-to-digital converters(ADCs) are assumed to equip at the base station(BS), is investigated in this pap...The uplink achievable rate of massive multiple-input-multiple-output(MIMO) systems, where the low-resolution analog-to-digital converters(ADCs) are assumed to equip at the base station(BS), is investigated in this paper. We assume that only imperfect channel station information is known at the BS. Then a new MMSE receiver is designed by taking not only the Gaussian noise, but also the channel estimation error and quantizer noise into account. By using the Stieltjes transform of random matrix, we further derive a tight asymptotic equivalent for the uplink achievable rate with proposed MMSE receiver. We present a detailed analysis for the number of BS antennas through the expression of the achievable rates and validate the results using numerical simulations. It is also shown that we can compensate the performance loss due to the low-resolution quantization by increasing the number of antennas at the BS.展开更多
Canonical correlation analysis ( CCA ) based methods for low-resolution ( LR ) face recognition involve face images with different resolutions ( or multi-resolutions ), i.e.LR and high-resolution ( HR ) .For single-re...Canonical correlation analysis ( CCA ) based methods for low-resolution ( LR ) face recognition involve face images with different resolutions ( or multi-resolutions ), i.e.LR and high-resolution ( HR ) .For single-resolution face recognition , researchers have shown that utilizing spatial information is beneficial to improving the recognition accuracy , mainly because the pixels of each face are not independent but spatially correlated.However , for a multi-resolution scenario , there are no related works.Therefore , a method named spatial regularization of canonical correlation analysis ( SRCCA ) is developed for LR face recognition to improve the performance of CCA by the regularization utilizing spatial information of different resolution faces.Furthermore , the impact of LR and HR spatial regularization terms on LR face recognition is analyzed through experiments.展开更多
Low-resolution(LR)fine-grained image recognition requires the ability to recognize the subcategories of LR samples with limited fine-grained details.The existing methods do not make full use of the guiding and constra...Low-resolution(LR)fine-grained image recognition requires the ability to recognize the subcategories of LR samples with limited fine-grained details.The existing methods do not make full use of the guiding and constraining capabilities of category-related knowledge to recover and extract the fine-grained features of LR data;thus these methods have a limited ability to learn the global and local fine-grained features of LR data.In this paper,we propose an enhanced feature representation network(EFR-Net)based on categorical knowledge guidance to capture delicate and reliable fine-grained feature descriptions of LR data and improve the recognition accuracy.First,to overcome the challenges posed by the limited fine-grained details in LR data,we design a classwise distillation loss.This loss function transfers the high-quality features of class-specific high-resolution(HR)samples into the feature learning of the same-category LR samples by using a memory bank.In this way,the global representation of LR images is closer to the meaningful and high-quality image features.Second,considering that fine-grained discriminative features are often hidden in object parts,we present a group of part queries to learn the positional information where the discriminative cues exist across all categories,and we then use the queries to decode diverse and discriminative part features.The global representation,in combination with the local discriminative features,creates more comprehensive and meaningful feature descriptions of the LR fine-grained objects,thus improving the recognition performance.Extensive comparison experiments on four LR datasets demonstrate the effectiveness of EFR-Net.展开更多
Virus image classification is a significant and challenging issue in both clinical virology and medical image processing.Due to the low-resolution virus images in the original dataset,there is tricky difficulty in ext...Virus image classification is a significant and challenging issue in both clinical virology and medical image processing.Due to the low-resolution virus images in the original dataset,there is tricky difficulty in extracting useful features from this kind of poor quality images adopting the traditional feature extraction methods.In this paper,we propose an effective and robust method,which eliminates the drawbacks of traditional local feature extraction methods and conducts latent local texture feature extraction thus to promote the accuracy of virus image classification.Firstly,the multi-scale principal component analysis(PCA)filters are learned from all original images.Then,it establishes a scale space for each PCA-filtered image by 2D Gaussian function.Finally,some typical feature descriptors are employed to extract texture features from all images,which include the original image and its filtered images by PCA and Gaussian filters.Aiming at the classification of low-resolution images,the proposed method solves the difficulty in extracting the essential feature from the original image and captures its latent and principal texture information from different perspectives in different filtered images.Experimental results show that the classification accuracy of the proposed method is much higher than state-of-the-art methods in the same low-resolution virus dataset,reaching 88.00%.展开更多
Existing multi-person reconstruction methods require the human bodies in the input image to occupy a considerable portion of the picture.However,low-resolution human objects are ubiquitous due to trade-offbetween the ...Existing multi-person reconstruction methods require the human bodies in the input image to occupy a considerable portion of the picture.However,low-resolution human objects are ubiquitous due to trade-offbetween the field of view and target distance given a limited camera resolution.In this paper,we propose an end-to-end multi-task framework for multi-person inference from a low-resolution image(MILI).To perceive more information from a low-resolution image,we use pair-wise images at high resolution and low resolution for training,and design a restoration network with a simple loss for better feature extraction from the low-resolution image.To address the occlusion problem in multi-person scenes,we propose an occlusion-aware mask prediction network to estimate the mask of each person during 3D mesh regression.Experimental results on both small-scale scenes and large-scale scenes demonstrate that our method outperforms the state-of-the-art methods both quantitatively and qualitatively.The code is available at http://cic.tju.edu.cn/faculty/likun/projects/MILI.展开更多
Multiple Object Tracking(MOT)is essential for applications such as autonomous driving,surveillance,and analytics;However,challenges such as occlusion,low-resolution imaging,and identity switches remain persistent.We p...Multiple Object Tracking(MOT)is essential for applications such as autonomous driving,surveillance,and analytics;However,challenges such as occlusion,low-resolution imaging,and identity switches remain persistent.We propose HAMOT,a hierarchical adaptive multi-object tracker that solves these challenges with a novel,unified framework.Unlike previous methods that rely on isolated components,HAMOT incorporates a Swin Transformer-based Adaptive Enhancement(STAE)module—comprising Scene-Adaptive Transformer Enhancement and Confidence-Adaptive Feature Refinement—to improve detection under low-visibility conditions.The hierarchical DynamicGraphNeuralNetworkwith TemporalAttention(DGNN-TA)models both short-and long-termassociations,and the Adaptive Unscented Kalman Filter with Gated Recurrent Unit(AUKF-GRU)ensures accurate motion prediction.The novel Graph-Based Density-Aware Clustering(GDAC)improves occlusion recovery by adapting to scene density,preserving identity integrity.This integrated approach enables adaptive responses to complex visual scenarios,Achieving exceptional performance across all evaluation metrics,including aHigher Order TrackingAccuracy(HOTA)of 67.05%,a Multiple Object Tracking Accuracy(MOTA)of 82.4%,an ID F1 Score(IDF1)of 83.1%,and a total of 1052 Identity Switches(IDSW)on theMOT17;66.61%HOTA,78.3%MOTA,82.1%IDF1,and a total of 748 IDSWonMOT20;and 66.4%HOTA,92.32%MOTA,and 68.96%IDF1 on DanceTrack.With fixed thresholds,the full HAMOT model(all six components)achieves real-time functionality at 24 FPS on MOT17 using RTX3090,ensuring robustness and scalability for real-world MOT applications.展开更多
Single image super-resolution has attracted increasing attention and has a wide range of applications in satellite imaging, medical imaging, computer vision, security surveillance imaging, remote sensing, objection de...Single image super-resolution has attracted increasing attention and has a wide range of applications in satellite imaging, medical imaging, computer vision, security surveillance imaging, remote sensing, objection detection, and recognition. Recently, deep learning techniques have emerged and blossomed, producing " the state-of-the-art” in many domains. Due to their capability in feature extraction and mapping, it is very helpful to predict high-frequency details lost in low-resolution images. In this paper, we give an overview of recent advances in deep learning-based models and methods that have been applied to single image super-resolution tasks. We also summarize, compare and discuss various models from the past and present for comprehensive understanding and finally provide open problems and possible directions for future research.展开更多
The Global Rainforest Mapping (GRFM) project was initiated in 1995 and, through a dedicated data acquisition policy by the National Space Development Agency of Japan (NASDA), data acquisitions could be completed withi...The Global Rainforest Mapping (GRFM) project was initiated in 1995 and, through a dedicated data acquisition policy by the National Space Development Agency of Japan (NASDA), data acquisitions could be completed within a 1.5-year period, resulting in a spatially and temporally homogeneous coverage to contain the entire Amazon Basin from the Atlantic to the Pacific; Central America up to the Yucatan Peninsular in Mexico; equatorial Africa from Madagascar and Kenya in the east to Sierra Leone in the west; and Southeast Asia, including Papua New Guinea. To some extent, GRFM project is an international endeavor led by NASDA, with the goal of producing spatially and temporally contiguous Synthetic Aperture Radar (SAR) data sets over the tropical belt on the Earth by use of the JERS-1 L-band SAR, through the generation of semi-continental, 100m resolution, image mosaics. The GRFM project relies on extensive collaboration with the National Aeronautics and Space Administration (NASA), the Joint Research Center of the European Commission (JRC) and the Japanese Ministry of International Trade and Industry (MITI) for data acquisition, processing, validation and product generation. A science program is underway in parallel with product generation. This involves the agencies mentioned above, as well as a large number of international organizations, universities and individuals to perform field activities and data analysis at different levels.展开更多
To achieve de novo protein structure determination of challenging cases, multi-wavelength anomalous diffraction(MAD) and multiple isomorphous replacement(MIR) phasing can be powerful tools to obtain low-resolution ini...To achieve de novo protein structure determination of challenging cases, multi-wavelength anomalous diffraction(MAD) and multiple isomorphous replacement(MIR) phasing can be powerful tools to obtain low-resolution initial phases from heavy-atom derivative datasets, then phase extension is needed against high-resolution data to obtain accurate structures.In this context, we propose a direct-methods procedure here that could improve the initial low-resolution MAD/MIR phase quality.And accordingly, an automated process for extending initial phases to high resolution is also described.These two procedures are both implanted in the newly released IPCAS pipeline.Three cases are used to perform the test, including one set of 4.17 ? MAD data from a membrane protein and two sets of MAD/MIR data with derivatives truncated down to 6.80 ? and 6.90 ?, respectively.All the results have shown that the initial phases generated from the direct-methods procedure are better than that from the conventional MAD/MIR methods.The automated phase extensions for the latter two cases starting from 6.80 ? to 3.00 ? and 6.90 ? to 2.80 ? are proved to be successful, leading to complete models.This may provide convenient and reliable tools for phase improvement and phase extension in difficult low-resolution tasks.展开更多
Machine learning methods, one type of methods used in artificial intelligence, are now widely used to analyze two-dimensional (2D) images in various fields. In these analyses, estimating the boundary between two regio...Machine learning methods, one type of methods used in artificial intelligence, are now widely used to analyze two-dimensional (2D) images in various fields. In these analyses, estimating the boundary between two regions is basic but important. If the model contains stochastic factors such as random observation errors, determining the boundary is not easy. When the probability distributions are mis-specified, ordinal methods such as probit and logit maximum likelihood estimators (MLE) have large biases. The grouping estimator is a semiparametric estimator based on the grouping of data that does not require specific probability distributions. For 2D images, the grouping is simple. Monte Carlo experiments show that the grouping estimator clearly improves the probit MLE in many cases. The grouping estimator essentially makes the resolution density lower, and the present findings imply that methods using low-resolution image analyses might not be the proper ones in high-density image analyses. It is necessary to combine and compare the results of high- and low-resolution image analyses. The grouping estimator may provide theoretical justifications for such analysis.展开更多
Low-resolution analog-to-digital converter(ADC)is a promising solution to reduce hardware cost and power consumption in generalized frequency division multiplexing(GFDM)systems.The severe nonlinear distortion of ADCs ...Low-resolution analog-to-digital converter(ADC)is a promising solution to reduce hardware cost and power consumption in generalized frequency division multiplexing(GFDM)systems.The severe nonlinear distortion of ADCs and the non-orthogonality of GFDM make receiver design a great challenge.In this paper,we propose a novel model-driven receiver architecture for GFDM with low-resolution ADCs.Orthogonal approximate message passing(OAMP)framework is combined with the classical linear estimator in this work to create a robust iterative receiver for GFDM systems with low-precision ADCs.The corresponding model-driven network is organized based on the proposed novel iterative algorithm according to the procedures of the receiver.The network of OAMP can reduce the gap between the approximate algorithm and the Bayesian optimal result due to the information loss of ADCs.The signal flow of the neural network is designed by unfolding the iterative algorithms for channel estimation and data detection.Numerical results are provided to show that the proposed OAMP-based receiver algorithm outperforms traditional receivers and the model-driven network can further improve the system performance on the basis of the corresponding novel algorithm.展开更多
Turbulence,a complex multi-scale phenomenon inherent in fluid flow systems,presents critical challenges and opportunities for understanding physical mechanisms across scientific and engineering domains.Although high-r...Turbulence,a complex multi-scale phenomenon inherent in fluid flow systems,presents critical challenges and opportunities for understanding physical mechanisms across scientific and engineering domains.Although high-resolution(HR)turbulence data remain indispensable for advancing both theoretical insights and engineering solutions,their acquisition is severely limited by prohibitively high computational costs.While deep learning architectures show transformative potential in reconstructing high-fidelity flow representations from sparse measurements,current methodologies suffer from two inherent constraints:strict reliance on perfectly paired training data and inability to perform multi-scale reconstruction within a unified framework.To address these challenges,we propose HADF,a hash-adaptive dynamic fusion implicit network for turbulence reconstruction.Specifically,we develop a low-resolution(LR)consistency loss that facilitates effective model training under conditions of missing paired data,eliminating the conventional requirement for fully matched LR and HR datasets.We further employ hash-adaptive spatial encoding and dynamic feature fusion to extract turbulence features,mapping them with implicit neural representations for reconstruction at arbitrary resolutions.Experimental results demonstrate that HADF achieves superior performance in global reconstruction accuracy and local physical properties compared to state-of-the-art models.It precisely recovers fine turbulence details for partially unpaired data conditions and diverse resolutions by training only once while maintaining robustness against noise.展开更多
基金supported by National Hi-tech Research and Development Program of China (863 Program,Grant No.2008AA11A126)Program for New Century Excellent Talents in University of China (Grant No. NCET-10-0498)
文摘Resolvers are normally employed for rotor positioning in motors for electric vehicles, but resolvers are expensive and vulnerable to vibrations. Hall sensors have the advantages of low cost and high reliability, but the positioning accuracy is low. Motors with Hall sensors are typically controlled by six-step commutation algorithm, which brings high torque ripple. This paper studies the high-performance driving and braking control of the in-wheel permanent magnetic synchronous motor （PMSM） based on low-resolution Hall sensors. Field oriented control （FOC） based on Hall-effect sensors is developed to reduce the torque ripple. The positioning accuracy of the Hall sensors is improved by interpolation between two consecutive Hall signals using the estimated motor speed. The position error from the misalignment of the Hall sensors is compensated by the precise calibration of Hall transition timing. The braking control algorithms based on six-step commutation and FOC are studied. Two variants of the six-step commutation braking control, namely, half-bridge commutation and full-bridge commutation, are discussed and compared, which shows that the full-bridge commutation could better explore the potential of the back electro-motive forces （EMF）, thus can deliver higher efficiency and smaller current ripple. The FOC braking is analyzed with the phasor diagrams. At a given motor speed, the motor turns from the regenerative braking mode into the plug braking mode if the braking torque exceeds a certain limit, which is proportional to the motor speed. Tests in the dynamometer show that a smooth control could be realized by FOC driving control and the highest efficiency and the smallest current ripple could be achieved by FOC braking control, compared to six-step commutation braking control. Therefore, FOC braking is selected as the braking control algorithm for electric vehicles. The proposed research ensures a good motor control performance while maintaining low cost and high reliability.
基金This study is supported by the Key Program of Chinese Academy of Sciences KZCX3 SW-221the National Natural Science Foundation of China(Grant No.40233033 and 40221503).
文摘A comparison study is performed to contrast the improvements in the tropical Pacific oceanic state of a low-resolution model respectively via data assimilation and by an increase in horizontal resolution. A low resolution model (LR) (1°lat by 2°lon) and a high-resolution model (HR) (0.5°lat by 0.5°lon) are employed for the comparison. The authors perform 20-yr numerical experiments and analyze the annual mean fields of temperature and salinity. The results indicate that the low-resolution model with data assimilation behaves better than the high-resolution model in the estimation of ocean large-scale features. From 1990 to 2000, the average of HR's RMSE (root-mean-square error) relative to independent Tropical Atmosphere Ocean project (TAO) mooring data at randomly selected points is 0.97℃ compared to a RMSE of 0.56℃ for LR with temperature assimilation. Moreover, the LR with data assimilation is more frugal in computation. Although there is room to improve the high-resolution model, the low-resolution model with data assimilation may be an advisable choice in achieving a more realistic large-scale state of the ocean at the limited level of information provided by the current observational system.
基金supported in part by the Key Research&Development Plan of Jiangsu Province(No.BE2018108)National Nature Science Foundation of China(Nos.61701198&61772243)+3 种基金Nature Science Foundation of Jiangsu Province(No.BK20170557)Nature Science Foundation for Higher Education Institutions of Jiangsu Province of China(No.17KJB510009)the open research fund of National Mobile Communications Research Laboratory,Southeast University(No.2018D13)Young Talent Project of Jiangsu University and Postgraduate Research&Practice Innovation Program of Jiangsu Province(No.SJCX18_0742)
文摘In a millimeter-wave(mmWave)Massive multiple-input multiple-output(MIMO)systems,full-digital beamforming(i.e.,connecting each antenna with a specific radio-frequency(RF)chain)becomes inefficient due to the hardware cost and power consumption.Therefore,hybrid analog and digital transceiver where the number of RF chains are much smaller than that of the antennas has drawn great research interest.In this work,we investigate the use of low-resolution analog-to-digital converters(ADCs)in the uplink of multi-user hybrid and full-digital mmWave Massive MIMO systems.To be specific,we compare the performance of full-digital minimum mean square error(MMSE)and hybrid MMSE beamforming in both sum rates and energy efficiency.Accurate approximations of sum rates and energy efficiency are provided for both schemes,which captures the dominant factors.The analytical results show that full-digital beamforming outperforms hybrid beamforming in terms of sum rates and requires only a small portion(γ)of antennas used by hybrid beamforming to achieve the same sum rates.We given sufficient condition for full-digital beamforming to outperform hybrid beamforming in terms of energy efficiency.Moreover,an algorithm is proposed to search for the optimal ADC resolution bits.Numerical results demonstrate the correctness of the analysis.
基金supported by National Natural Science Foundation of China(Nos.61271432 and 61333016)
文摘In this paper, we focus on low-resolution human detection and propose a partial least squares-canonical correlation analysis （PLS-CCA） for outdoor video surveillance. The analysis relies on heterogeneous features fusion-based human detection method. The proposed method can not only explore the relation between two individual heterogeneous features as much as possible, but also can robustly describe the visual appearance of humans with complementary information. Compared with some other methods, the experimental results show that the proposed method is effective and has a high accuracy, precision, recall rate and area under curve （AUC） value at the same time, and offers a discriminative and stable recognition performance.
文摘IRAS low-resolution spectra are presented for 36 HAEBE stars. It is found that silicate dust in amorphous or glassy form is common material in the circumstellar disks or/and shells of HAEBE stars. It is also found that the PAH feature is often appeared as well.
基金supported by the National High Technology Research and Development Program of China(863 Program)(2009AA011204)
文摘Nonparametric time-of-arrival（TOA） estimators for impulse radio ultra-wideband（IR-UWB） signals are proposed. Nonparametric detection is obviously useful in situations where detailed information about the statistics of the noise is unavailable or not accurate. Such TOA estimators are obtained based on conditional statistical tests with only a symmetry distribution assumption on the noise probability density function. The nonparametric estimators are attractive choices for low-resolution IR-UWB digital receivers which can be implemented by fast comparators or high sampling rate low resolution analog-to-digital converters（ADCs）,in place of high sampling rate high resolution ADCs which may not be available in practice. Simulation results demonstrate that nonparametric TOA estimators provide more effective and robust performance than typical energy detection（ED） based estimators.
基金supported by the Beijing Natural Science Foundation under Grant No. L172030the Beijing Municipal Natural Science Foundation under Grant No. 4174102+2 种基金NSFC Project under Grants No. 61471027the National Natural Science Foundation of China under Grant No. 61701017 and Grant No. 61601018the Open Research Fund through the National Mobile Communications Research Laboratory, Southeast University, under Grant No. 2017D01
文摘The uplink achievable rate of massive multiple-input-multiple-output(MIMO) systems, where the low-resolution analog-to-digital converters(ADCs) are assumed to equip at the base station(BS), is investigated in this paper. We assume that only imperfect channel station information is known at the BS. Then a new MMSE receiver is designed by taking not only the Gaussian noise, but also the channel estimation error and quantizer noise into account. By using the Stieltjes transform of random matrix, we further derive a tight asymptotic equivalent for the uplink achievable rate with proposed MMSE receiver. We present a detailed analysis for the number of BS antennas through the expression of the achievable rates and validate the results using numerical simulations. It is also shown that we can compensate the performance loss due to the low-resolution quantization by increasing the number of antennas at the BS.
基金Supported by the National Natural Science Foundation of China(6117015161070133+2 种基金60903130)the Natural Science Research Project of Higher Education of Jiangsu Province(12KJB520018)the Research Foundation of Nanjing University of Aeronautics and Astronautics(NP2011030)
文摘Canonical correlation analysis ( CCA ) based methods for low-resolution ( LR ) face recognition involve face images with different resolutions ( or multi-resolutions ), i.e.LR and high-resolution ( HR ) .For single-resolution face recognition , researchers have shown that utilizing spatial information is beneficial to improving the recognition accuracy , mainly because the pixels of each face are not independent but spatially correlated.However , for a multi-resolution scenario , there are no related works.Therefore , a method named spatial regularization of canonical correlation analysis ( SRCCA ) is developed for LR face recognition to improve the performance of CCA by the regularization utilizing spatial information of different resolution faces.Furthermore , the impact of LR and HR spatial regularization terms on LR face recognition is analyzed through experiments.
基金supported by the Dalian Youth Science and Technology Star Program under Grant No.2023RQ014the Basic Education Project of Liaoning Province of China under Grant No.JYTQN2023101+2 种基金the Interdisciplinary Project of Dalian University under Grant No.DLUXK-2023-QN-016the Support Plan for Key Field Innovation Team of Dalian under Grant No.2021RT06the 111 Project of China under Grant No.D23006.
文摘Low-resolution(LR)fine-grained image recognition requires the ability to recognize the subcategories of LR samples with limited fine-grained details.The existing methods do not make full use of the guiding and constraining capabilities of category-related knowledge to recover and extract the fine-grained features of LR data;thus these methods have a limited ability to learn the global and local fine-grained features of LR data.In this paper,we propose an enhanced feature representation network(EFR-Net)based on categorical knowledge guidance to capture delicate and reliable fine-grained feature descriptions of LR data and improve the recognition accuracy.First,to overcome the challenges posed by the limited fine-grained details in LR data,we design a classwise distillation loss.This loss function transfers the high-quality features of class-specific high-resolution(HR)samples into the feature learning of the same-category LR samples by using a memory bank.In this way,the global representation of LR images is closer to the meaningful and high-quality image features.Second,considering that fine-grained discriminative features are often hidden in object parts,we present a group of part queries to learn the positional information where the discriminative cues exist across all categories,and we then use the queries to decode diverse and discriminative part features.The global representation,in combination with the local discriminative features,creates more comprehensive and meaningful feature descriptions of the LR fine-grained objects,thus improving the recognition performance.Extensive comparison experiments on four LR datasets demonstrate the effectiveness of EFR-Net.
基金The research was supported by the National Natural Science Foundation of China(Nos.11471208,11472073,61772104,11701357,11771276).
文摘Virus image classification is a significant and challenging issue in both clinical virology and medical image processing.Due to the low-resolution virus images in the original dataset,there is tricky difficulty in extracting useful features from this kind of poor quality images adopting the traditional feature extraction methods.In this paper,we propose an effective and robust method,which eliminates the drawbacks of traditional local feature extraction methods and conducts latent local texture feature extraction thus to promote the accuracy of virus image classification.Firstly,the multi-scale principal component analysis(PCA)filters are learned from all original images.Then,it establishes a scale space for each PCA-filtered image by 2D Gaussian function.Finally,some typical feature descriptors are employed to extract texture features from all images,which include the original image and its filtered images by PCA and Gaussian filters.Aiming at the classification of low-resolution images,the proposed method solves the difficulty in extracting the essential feature from the original image and captures its latent and principal texture information from different perspectives in different filtered images.Experimental results show that the classification accuracy of the proposed method is much higher than state-of-the-art methods in the same low-resolution virus dataset,reaching 88.00%.
基金partly supported by the National Natural Science Foundation of China(62122058,62171317,and 62231018).
文摘Existing multi-person reconstruction methods require the human bodies in the input image to occupy a considerable portion of the picture.However,low-resolution human objects are ubiquitous due to trade-offbetween the field of view and target distance given a limited camera resolution.In this paper,we propose an end-to-end multi-task framework for multi-person inference from a low-resolution image(MILI).To perceive more information from a low-resolution image,we use pair-wise images at high resolution and low resolution for training,and design a restoration network with a simple loss for better feature extraction from the low-resolution image.To address the occlusion problem in multi-person scenes,we propose an occlusion-aware mask prediction network to estimate the mask of each person during 3D mesh regression.Experimental results on both small-scale scenes and large-scale scenes demonstrate that our method outperforms the state-of-the-art methods both quantitatively and qualitatively.The code is available at http://cic.tju.edu.cn/faculty/likun/projects/MILI.
基金supported in part by Multimedia University under the Research Fellow Grant MMUI/250008in part by Telekom Research&Development Sdn Bhd under Grants RDTC/241149 and RDTC/231095+1 种基金Princess Nourah bint Abdulrahman University Researchers Supporting Project number(PNURSP2025R140)Princess Nourah bint Abdulrahman University,Riyadh,Saudi Arabia.
文摘Multiple Object Tracking(MOT)is essential for applications such as autonomous driving,surveillance,and analytics;However,challenges such as occlusion,low-resolution imaging,and identity switches remain persistent.We propose HAMOT,a hierarchical adaptive multi-object tracker that solves these challenges with a novel,unified framework.Unlike previous methods that rely on isolated components,HAMOT incorporates a Swin Transformer-based Adaptive Enhancement(STAE)module—comprising Scene-Adaptive Transformer Enhancement and Confidence-Adaptive Feature Refinement—to improve detection under low-visibility conditions.The hierarchical DynamicGraphNeuralNetworkwith TemporalAttention(DGNN-TA)models both short-and long-termassociations,and the Adaptive Unscented Kalman Filter with Gated Recurrent Unit(AUKF-GRU)ensures accurate motion prediction.The novel Graph-Based Density-Aware Clustering(GDAC)improves occlusion recovery by adapting to scene density,preserving identity integrity.This integrated approach enables adaptive responses to complex visual scenarios,Achieving exceptional performance across all evaluation metrics,including aHigher Order TrackingAccuracy(HOTA)of 67.05%,a Multiple Object Tracking Accuracy(MOTA)of 82.4%,an ID F1 Score(IDF1)of 83.1%,and a total of 1052 Identity Switches(IDSW)on theMOT17;66.61%HOTA,78.3%MOTA,82.1%IDF1,and a total of 748 IDSWonMOT20;and 66.4%HOTA,92.32%MOTA,and 68.96%IDF1 on DanceTrack.With fixed thresholds,the full HAMOT model(all six components)achieves real-time functionality at 24 FPS on MOT17 using RTX3090,ensuring robustness and scalability for real-world MOT applications.
基金the support from the Shanxi Hundred People Plan of China
文摘Single image super-resolution has attracted increasing attention and has a wide range of applications in satellite imaging, medical imaging, computer vision, security surveillance imaging, remote sensing, objection detection, and recognition. Recently, deep learning techniques have emerged and blossomed, producing " the state-of-the-art” in many domains. Due to their capability in feature extraction and mapping, it is very helpful to predict high-frequency details lost in low-resolution images. In this paper, we give an overview of recent advances in deep learning-based models and methods that have been applied to single image super-resolution tasks. We also summarize, compare and discuss various models from the past and present for comprehensive understanding and finally provide open problems and possible directions for future research.
基金Knowledge Innovation Project of CAS,No. KZCX02-308
文摘The Global Rainforest Mapping (GRFM) project was initiated in 1995 and, through a dedicated data acquisition policy by the National Space Development Agency of Japan (NASDA), data acquisitions could be completed within a 1.5-year period, resulting in a spatially and temporally homogeneous coverage to contain the entire Amazon Basin from the Atlantic to the Pacific; Central America up to the Yucatan Peninsular in Mexico; equatorial Africa from Madagascar and Kenya in the east to Sierra Leone in the west; and Southeast Asia, including Papua New Guinea. To some extent, GRFM project is an international endeavor led by NASDA, with the goal of producing spatially and temporally contiguous Synthetic Aperture Radar (SAR) data sets over the tropical belt on the Earth by use of the JERS-1 L-band SAR, through the generation of semi-continental, 100m resolution, image mosaics. The GRFM project relies on extensive collaboration with the National Aeronautics and Space Administration (NASA), the Joint Research Center of the European Commission (JRC) and the Japanese Ministry of International Trade and Industry (MITI) for data acquisition, processing, validation and product generation. A science program is underway in parallel with product generation. This involves the agencies mentioned above, as well as a large number of international organizations, universities and individuals to perform field activities and data analysis at different levels.
基金Project supported by the National Basic Research Program of China(Grant No.2011CB911100)of the Ministry of Science and Technology of China
文摘To achieve de novo protein structure determination of challenging cases, multi-wavelength anomalous diffraction(MAD) and multiple isomorphous replacement(MIR) phasing can be powerful tools to obtain low-resolution initial phases from heavy-atom derivative datasets, then phase extension is needed against high-resolution data to obtain accurate structures.In this context, we propose a direct-methods procedure here that could improve the initial low-resolution MAD/MIR phase quality.And accordingly, an automated process for extending initial phases to high resolution is also described.These two procedures are both implanted in the newly released IPCAS pipeline.Three cases are used to perform the test, including one set of 4.17 ? MAD data from a membrane protein and two sets of MAD/MIR data with derivatives truncated down to 6.80 ? and 6.90 ?, respectively.All the results have shown that the initial phases generated from the direct-methods procedure are better than that from the conventional MAD/MIR methods.The automated phase extensions for the latter two cases starting from 6.80 ? to 3.00 ? and 6.90 ? to 2.80 ? are proved to be successful, leading to complete models.This may provide convenient and reliable tools for phase improvement and phase extension in difficult low-resolution tasks.
文摘Machine learning methods, one type of methods used in artificial intelligence, are now widely used to analyze two-dimensional (2D) images in various fields. In these analyses, estimating the boundary between two regions is basic but important. If the model contains stochastic factors such as random observation errors, determining the boundary is not easy. When the probability distributions are mis-specified, ordinal methods such as probit and logit maximum likelihood estimators (MLE) have large biases. The grouping estimator is a semiparametric estimator based on the grouping of data that does not require specific probability distributions. For 2D images, the grouping is simple. Monte Carlo experiments show that the grouping estimator clearly improves the probit MLE in many cases. The grouping estimator essentially makes the resolution density lower, and the present findings imply that methods using low-resolution image analyses might not be the proper ones in high-density image analyses. It is necessary to combine and compare the results of high- and low-resolution image analyses. The grouping estimator may provide theoretical justifications for such analysis.
基金This work was supported in part by the National Key Research and Development Program(2018YFA0701602)the National Natural Science Foundation of China for Distinguished Young Scholars of China(Nos.61625106,61531011)+1 种基金The work of C.K.Wen was supported in part by the Ministry of Science and Technology of Taiwan(MOST 106-2221-E-110-019)the ITRI in Hsinchu,Taiwan,China。
文摘Low-resolution analog-to-digital converter(ADC)is a promising solution to reduce hardware cost and power consumption in generalized frequency division multiplexing(GFDM)systems.The severe nonlinear distortion of ADCs and the non-orthogonality of GFDM make receiver design a great challenge.In this paper,we propose a novel model-driven receiver architecture for GFDM with low-resolution ADCs.Orthogonal approximate message passing(OAMP)framework is combined with the classical linear estimator in this work to create a robust iterative receiver for GFDM systems with low-precision ADCs.The corresponding model-driven network is organized based on the proposed novel iterative algorithm according to the procedures of the receiver.The network of OAMP can reduce the gap between the approximate algorithm and the Bayesian optimal result due to the information loss of ADCs.The signal flow of the neural network is designed by unfolding the iterative algorithms for channel estimation and data detection.Numerical results are provided to show that the proposed OAMP-based receiver algorithm outperforms traditional receivers and the model-driven network can further improve the system performance on the basis of the corresponding novel algorithm.
基金Project supported by the National Natural Science Foundation of China(No.12402349)the Natural Science Foundation of Hunan Province(No.2024JJ6468)+1 种基金the Youth Foundation of the National University of Defense Technology(No.ZK2023-11)the National Key Research and Development Program of China(No.2021YFB0300101)。
文摘Turbulence,a complex multi-scale phenomenon inherent in fluid flow systems,presents critical challenges and opportunities for understanding physical mechanisms across scientific and engineering domains.Although high-resolution(HR)turbulence data remain indispensable for advancing both theoretical insights and engineering solutions,their acquisition is severely limited by prohibitively high computational costs.While deep learning architectures show transformative potential in reconstructing high-fidelity flow representations from sparse measurements,current methodologies suffer from two inherent constraints:strict reliance on perfectly paired training data and inability to perform multi-scale reconstruction within a unified framework.To address these challenges,we propose HADF,a hash-adaptive dynamic fusion implicit network for turbulence reconstruction.Specifically,we develop a low-resolution(LR)consistency loss that facilitates effective model training under conditions of missing paired data,eliminating the conventional requirement for fully matched LR and HR datasets.We further employ hash-adaptive spatial encoding and dynamic feature fusion to extract turbulence features,mapping them with implicit neural representations for reconstruction at arbitrary resolutions.Experimental results demonstrate that HADF achieves superior performance in global reconstruction accuracy and local physical properties compared to state-of-the-art models.It precisely recovers fine turbulence details for partially unpaired data conditions and diverse resolutions by training only once while maintaining robustness against noise.
