Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Yuan Yan TangSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works290 from public data
- Multiscale patch-based contrast measure for small infrared target detection780
Experimental results on three sequences demonstrate that the proposed small target detection method can not only suppress background clutter effectively even if with strong noise interference, but also detect targets accurately with low false alarm rate and high speed.
- Segmentation of retinal blood vessels using the radial projection and semi-supervised approach402
A novel scheme of extracting the retinal vessels based on the radial projection and semi-supervised method that can achieve improved detection of thin vessels and decrease false detection of vessels in pathological regions compared to rival solutions is presented.
- Hierarchical Bilinear Pooling for Fine-Grained Visual Recognition353
A cross-layer bilinear pooling approach is proposed to capture the inter-layer part feature relations, which results in superior performance compared with other bilinears pooling based approaches.
- 249
- FREE: Feature Refinement for Generalized Zero-Shot Learning229
A feature refinement module that in-corporates semantic→visual mapping into a unified generative model to refine the visual features of seen and unseen class samples, and a self-adaptive margin center loss (SAMC-loss) that cooperates with a semantic cycle-consistency loss to guide FR to learn class- and semantically-relevant representations.
- 207
- TransZero: Attribute-Guided Transformer for Zero-Shot Learning206
This paper proposes an attribute-guided Transformer network to learn the attribute localization for discriminative visual-semantic embedding representations in ZSL, termed TransZero, which achieves a new state-of-the-art on three ZSL benchmarks.
- Shape matching and classification using height functions193
A novel shape descriptor for matching and recognizing 2D object silhouettes that is not only invariant to geometric transformations such as translation, rotation and scaling but also insensitive to nonlinear deformations due to noise and occlusion is proposed.
- Robust face recognition via occlusion dictionary learning180
The proposed novel structured dictionary learning method achieves better results than the existing sparse representation based face recognition methods, especially in dealing with large region contiguous occlusion and severe illumination variation, while the computational cost is much lower.
- MSDN: Mutually Semantic Distillation Network for Zero-Shot Learning163
A Mutually Semantic Distillation Network (MSDN) is proposed, which progressively distills the intrinsic semantic representations between visual and attribute features for ZSL, leading to new state-of-the-art performances on three popular challenging benchmarks.
- 148
- 145
- Efficient single image dehazing and denoising: An efficient multi-scale correlated wavelet approach136
This paper presents an efficient multi-scale correlated wavelet approach to solve the image dehazing and denoising problem in the frequency domain and finds a generic regularity in nature images that the haze is typically distributed in the low frequency spectrum of its multi- scale wavelet decomposition.
- Connected Component Model for Multi-Object Tracking133
This paper shows that the relation between associations of two observations is the equivalence relation in the data association problem, based on the spatial-temporal constraint that the trajectories of different objects must be disjoint, and develops a connected component model (CCM), which can efficiently obtain the global solution of the MDA problem for multi-object tracking by optimizing a sequence of independent data association subproblems.
- Multi-view low-rank dictionary learning for image classification131
This paper provides a multi-view dictionary low-rank regularization term to solve the noise problem, and designs a structural incoherence constraint for multi-View DL, such that redundancy among dictionaries of different views can be reduced.
- Automatic kidney segmentation in ultrasound images using subsequent boundary distance regression and pixelwise classification networks121
This study first uses deep neural networks pre-trained for classification of natural images to extract high-level image features from US images to segment the kidneys automatically, and adopts a data-augmentation method based on kidney shape registration to generate enriched training data from a small number of US images.
- Writer identification of Chinese handwriting documents using hidden Markov tree model120
A novel method based on hidden Markov tree (HMT) model in wavelet domain for off-line, text-independent writer identification of Chinese handwriting documents is presented and experiments show this HMT method not only achieves better identification results but also greatly reduces the elapsed time on computation.
- A new weighted mean filter with a two-phase detector for removing impulse noise110
Experimental results show that the proposed method outperforms all the tested state-of-the-art denoising methods with respect to the visual effects and quantitative measure results.
- Skeletonization of ribbon-like shapes based on a new wavelet function108
Experimental results show that the proposed wavelet-based scheme is capable of extracting exactly the skeleton of the Ribbon-like shape with different width as well as different gray-levels, and is robust against noise and affine transformation.
- 106
- View Vertically: A Hierarchical Network for Trajectory Prediction via Fourier Spectrums104
This work proposes a hierarchical network V$^2-Net, which contains two sub-networks, to hierarchically model and predict agents' trajectories with trajectory spectrums, and shows the competitiveness and superiority of V$2$-Net on both ETH-UCY benchmark and the Stanford Drone Dataset.
- TransZero++: Cross Attribute-Guided Transformer for Zero-Shot Learning99
A cross attribute-guided Transformer network to refine visual features and learn accurate attribute localization for key semantic knowledge representations in ZSL, which achieves the new state-of-the-art results on three golden ZSL benchmarks and on the large-scale ImageNet dataset.
- 96
- 92
- 91
- An adaptive hybrid pattern for noise-robust texture analysis88
A noise-robust adaptive hybrid pattern (AHP) for noised texture analysis that achieves state-of-the-art performance in noisy face recognition and an adaptive quantization algorithm designed for robust angular space quantization.
- Multi-view manifold learning with locality alignment85
A multi-view manifold learning with locality alignment (MVML-LA) framework to learn a common yet discriminative low-dimensional latent space that contain sufficient information of original inputs is proposed.
- 84
- 84
- Few-shot object detection: Research advances and challenges74
A novel FSOD taxonomy method is proposed and the plentifully remarkable FSOD algorithms are surveyed based on this fact to report a comprehensive overview that facilitates a deeper understanding of the FSOD problem and the development of innovative solutions.
- Sparse discriminative multi-manifold embedding for one-sample face identification71
This paper proposes a two-step scheme for the OSPP problem by posing it as a representation and matching problem, and presents a novel manifold embedding algorithm, namely sparse discriminative multi-manifold embedding (SDMME), to learn the intrinsic representation beneath the raw data.
- CSCNet: Contextual semantic consistency network for trajectory prediction in crowded spaces70
A well-designed context-aware transfer is utilized to obtain the intermediate representations from the scene images and trajectories and eliminates the differences between social and physical interactions by aligning activity semantics and scene semantics to cross the Gap.
- Multi-view common component discriminant analysis for cross-view classification69
This work proposes the Multi-view Common Component Discriminant Analysis (MvCCDA), which incorporates supervised information and local geometric information into the common component extraction process to learn a discriminant common subspace and to discover the nonlinear structure embedded in multi-view data.
- 69
- SocialCircle: Learning the Angle-based Social Interaction Representation for Pedestrian Trajectory Prediction68
A new angle-based trainable social interaction representation, named SocialCircle, for continuously reflecting the context of social interactions at different angular orientations relative to the target agent, which quantitatively improves the prediction performance and qualitatively helps better simulate social interactions when forecasting pedestrian trajectories in a way that is consistent with human intuitions.
- ProClusEnsem: Predicting membrane protein types by fusing different modes of pseudo amino acid composition63
The experimental results on a benchmark dataset demonstrate the feasibility and effectiveness of the proposed algorithm named ProClusEnsem, which learns local metrics which are optimized for local subset of the whole proteins.
- Texture image retrieval based on non-tensor product wavelet filter banks63
A novel method is presented, which uses non-separable wavelet filter banks, to extract the features of texture images for texture image retrieval and can achieve better retrieval accuracies than db wavelets.
- Writer identification using global wavelet-based features63
A new method by combining wavelet transform and GGD model for writer identification of Chinese handwriting document is proposed and achieves a satisfied identification result and computational efficiency as well.
- Equally-Guided Discriminative Hashing for Cross-modal Retrieval59
This work discovers the connection between semantic structure preserving and discriminative methods, and directly encode multi-label annotations that act as high-level semantic features to build a common semantic structure preserves classifier.
- CDE-GAN: Cooperative Dual Evolution-Based Generative Adversarial Network58
Extensive experiments demonstrate that the proposed CDE-GAN achieves the competitive and superior performance in generating good quality and diverse samples over baselines, and to improve generative performance.
- 58
- A novel iris segmentation using radial-suppression edge detection57
A new approach for iris segmentation based on radial-suppression edge detection and a self-adaptive method of iris boundary detection is proposed to produce final iris boundaries.
- 56
- Recent Advances in Concept Drift Adaptation Methods for Deep Learning53
This paper mathematically describes the categories of concept drift including abrupt drift, gradual drift, recurrent drift, incremental drift, and divides existing studies into two categories, and analyzes the pros and cons of representative methods in each category.
- 53
- 52
- Visual-Augmented Dynamic Semantic Prototype for Generative Zero-Shot Learning51
This work proposes a novel Visual-Augmented Dynamic Semantic prototype method (termed VADS) to boost the generator to learn accurate semantic-visual mapping by fully exploiting the visual-augmented knowledge into semantic conditions.
- Writer identification using fractal dimension of wavelet subbands in gabor domain51
This paper proposes one new method for off-line, text-independent writer identification by using the fractal dimension of wavelet subbands in Gabor domain of the handwriting images to extract more effective features to distinguish the handwritings, and hence achieve much better identification results.
- 50
- Single object tracking via robust combination of particle filter and sparse representation50
This paper proposes a tracking method based on the robust combination of particle filter and reverse sparse representation (RC-PFRSR) to reduce the drifting and proposes a diagonal matrix α, whose diagonal line includes each patch contribution factor, to function each patch coefficient value of one candidate obtained by sparse representation.
- 50
- FACIAL BIOMETRICS USING NONTENSOR PRODUCT WAVELET AND 2D DISCRIMINANT TECHNIQUES44
A new facial biometric scheme is proposed in this paper and three steps are included, including a new nontensor product bivariate wavelet utilized to get different facial frequency components.
- Evolving Semantic Prototype Improves Generative Zero-Shot Learning42
A dynamic semantic prototype evolving (DSP) method is proposed to align the empirically predefined semantic prototypes and the real prototypes for class-related feature synthesis and indicates that evolving semantic prototype explores a virgin field in ZSL.
- Rotation invariant iris feature extraction using Gaussian Markov random fields with non-separable wavelet42
A new rotation invariant approach for iris feature extraction based on the non-separable wavelet is proposed in this paper and has a low error rate and achieves true rotation invariance.
- 42
- MSN: Multi-Style Network for Trajectory Prediction41
The Multi-Style Network (MSN) is proposed, which utilizes style proposal and stylized prediction using two sub-networks, to provide multi-style predictions in a novel categorical way adaptively to adaptively learn multiple diverse behavior styles through these channels.
- Distance learning by mining hard and easy negative samples for person re-identification41
A Hard and Easy Negative samples mining based Distance learning (HEND) approach for person re-identification is proposed, which learns the distance metric by designing different objective functions for hard and easy negative samples, such that the discriminative information contained in negative samples can be exploited more effectively.
- 41
- Multi-instance Deep Learning with Graph Convolutional Neural Networks for Diagnosis of Kidney Diseases Using Ultrasound Imaging40
A novel multi- instance deep learning method to build a robust classifier by treating multiple 2D US images of each individual subject as multiple instances of one bag, with better performance than alternative state-of-the-art multi-instance deep learning methods.
- 40
- Multi-instance Deep Learning of Ultrasound Imaging Data for Pattern Classification of Congenital Abnormalities of the Kidney and Urinary Tract in Children39
The multi-instance deep learning method provides an automatic and accurate means to extract informative features from ultrasound images and discriminate infants with PUV from male children with unilateral hydronephrosis.
- 39
- Generalization performance of magnitude-preserving semi-supervised ranking with graph-based regularization39
This work proposes a novel graph-based regularized algorithm which learns the ranking function in the semi-supervised learning framework which can exploit geometry of the data while preserving the magnitude of the preferences.
- 38
- 38
- 37
- Fully-Automatic Segmentation Of Kidneys In Clinical Ultrasound Images Using A Boundary Distance Regression Network36
A novel boundary distance regression deep neural network is developed to segment the kidneys, informed by the fact that the kidney boundaries are relatively consistent across images in terms of their appearance.
- Semantics-Conditioned Generative Zero-Shot Learning via Feature Refinement34
This paper introduces a visual feature refinement method (ViFR) to mitigate cross-dataset bias and advance generative ZSL, and proposes a self-adaptive margin center loss (SAMC-loss) that collaborates with semantic cycle-consistency loss to guide Post-FR in learning class- and semantically-relevant representations.
- EGANS: Evolutionary Generative Adversarial Network Search for Zero-Shot Learning34
Extensive experiments show that EGANS consistently improve existing generative ZSL methods on the standard CUB, SUN, AWA2 and FLO datasets, and indicate that the evolutionary NAS explores a virgin field in ZSL.
- 34
- 34
- Semantic Compression Embedding for Generative Zero-Shot Learning32
A novel Semantic Compression Embedding Guided Generation (SC-EGG) model, which cascades a semantic compression embedding network (SCEN) and an embedding guided generative network (EGGN) and learns a mapping from the class-level semantic space to the dense-semantic visual space, thus improving the discriminability of the synthesized dense- Semantic unseen visual features.
- Mixed Noise Removal via Robust Constrained Sparse Representation32
A robust constrained sparse representation (RCSR) method to remove mixed noise by using the center coefficient of similar patches as the guider which is approximated by the coefficient of query patch in sparse coding, so that the geometric structure of data can be well preserved.
- Regularized multivariate scatter correction32
This paper proposes regularized multivariate scatter correction (RMSC), which has taken nonlinear components into MSC model as well as regularization function for the weight vector w by using gradient projection sparse representation.
- A New Algorithm for Shoreline Extraction from Satellite Imagery with Non-Separable Wavelet and Level Set Method32
Results show that the proposed algorithm is applicable to satellite imagery, and the shoreline is robust to noises as well as blurring.
- Norm-guided Adaptive Visual Embedding for Zero-Shot Sketch-Based Image Retrieval31
A novel Norm-guided Adaptive Visual Embedding model, for adaptively building the common space based on visual similarity instead of language-based pre-defined prototypes, is proposed, which demonstrates the superiority of the NAVE over state-of-the-art competitors.
- A local region based approach to lip tracking31
A local region based approach to lip tracking, which consists of two phases: lip contour extraction for the first lip frame, and followed by lip tracking in the subsequent lip frames, which is efficient in comparison with the existing methods.
- STFT-like time frequency representations of nonstationary signal with arbitrary sampling schemes29
An alternative approach, namely short-time sparse learning via iterative minimization (ST-SLIM), which can provide sparser and slightly better TFR performance than its ST-IAA counterpart and extend the applicability of ST- IAA to signals in the missing data case is presented.
- 28
- 28
- Arabic font recognition based on diacritics features28
This paper presents the diacritics as the thumb of Arabic fonts which can be used individually to identify and recognize the font type and it can be integrated with OCR systems very easily.
- MelTrans: Mel-Spectrogram Relationship-Learning for Speech Emotion Recognition via Transformers27
A Transformer-based model called MelTrans, designed to distill critical clues from speech data by learning core features and long-range dependencies, is introduced, which highlights MelTrans’s ability to capture critical cues and long-range dependencies in speech data.
- ECEA: Extensible Co-Existing Attention for Few-Shot Object Detection27
An Extensible Co-Existing Attention (ECEA) module is proposed to enable the model to infer the global object according to the local parts and can assist the few-shot model to quickly adapt in extending local regions to co-existing regions.
- 26
- One global optimization method in network flow model for multiple object tracking26
This paper addresses the task of automatically tracking a variable number of objects in the scene of a monocular and uncalibrated camera, and proposes a global optimization method in network flow model for multiple object tracking.
- THINNING CHARACTER USING MODULUS MINIMA OF WAVELET TRANSFORM26
An efficient algorithm is proposed to extract visually satisfactory skeleton characteristics of the character using a convolutional neural network.
- 25
- CTR-Driven Advertising Image Generation with Multimodal Large Language Models24
A novel reward model is proposed to fine-tune pre-trained MLLMs through Reinforcement Learning (RL), which can jointly utilize multimodal features and accurately reflect user click preferences, enhancing the overall relevance and effectiveness of the advertising images.
- 24
- Causal Visual-semantic Correlation for Zero-shot Learning23
This work proposes a novel method termed Causal Visual-semantic Correlation (CVsC) to learn substantive visual-semantic correlation for ZSL, and designs a Conditional Correlation Loss to properly utilize semantic information as supervision for establishing visual-semantic correlation.
- Dynamic texture modeling and synthesis using multi-kernel Gaussian process dynamic model23
A Bayesian-based nonlinear dynamic texture modeling method for dynamic texture synthesis and a multi-kernel based Gaussian process dynamic model for latent dynamic behavior modeling are proposed.
- A novel joint tracker based on occlusion detection23
This paper proposes a new mechanism which can predict occlusion accurately and sensitively with MIL and SVM classifiers and proposes a classification-based template updating method, in which the templates are divided into two groups according to occlusions information and use opposite probability distributions to update the two groups.
- Deep medical cross-modal attention hashing22
A novel end-to-end deep hashing method, called Deep Medical Cross-Modal Attention Hashing (DMCAH), which extracts the global features utilizing global average pooling and local features by recurrent attention and reduces the semantic gap.
- 20
- 20
- 19
- 19
- Fast pedestrian detection based on region of interest and multi-block local binary pattern descriptors19
Experimental results showed that the proposed-model performs better than some state-of-the-art approaches, with suitable processing time for further operations such as tracking and imminent danger estimation.
- 19
- 18
- 18
- 18
- A method using long digital straight segments for fingerprint recognition18
A new method using long digital straight segments (LDSSs) for fingerprint recognition based on such a discovery that LDSSs in fingerprints can accurately characterize the global structure of fingerprints is proposed.
- Filter Pruning Based on Information Capacity and Independence17
A new filter pruning method that selects filters in an interpretable, multiperspective, and lightweight manner and can identify and prune the least important filters with less computation cost is introduced.
- Coarse-to-fine salient object detection with low-rank matrix recovery17
A novel and efficient LRMR-based saliency detection model under a coarse-to-fine framework that maps image features to refined saliency values, to significantly sharpen the object boundaries and to preserve the object entirety is introduced.
- Multiview Hybrid Embedding: A Divide-and-Conquer Approach17
This work proposes multiview hybrid embedding (MvHE), a unique method of dividing the problem of cross-view classification into three subproblems and building one model for each subproblem, designed to remove view discrepancy and discover the intrinsic nonlinear structure.
- 17
- Robust Visual Tracking Using Multi-Frame Multi-Feature Joint Modeling16
A novel correlation filter-based tracker is proposed, in which the temporal relatedness is reconciled under a multi-task learning framework and the multiple feature cues are modeled using aMulti-view learning approach, and it is demonstrated that the resulting regression model can be efficiently learned by exploiting the structure of blockwise diagonal matrix.
- 16
- HANDWRITING-BASED PERSONAL IDENTIFICATION16
A novel wavelet-based Generalized Gaussian Density (GGD) method for offline writer identification that not only achieves a better identification accuracy but also greatly reduces the elapsed time on calculation in the authors' experiments.
- 15
- Variance constrained partial least squares15
This paper derived the mathematical formulation of the relationship between mean square error (MSE) and these three factors and found that MSE is not monotonous with the product of the three factors.
- 14
- Unsupervised multiplex graph diffusion networks with multi-level canonical correlation analysis for multiplex graph representation learning13
This paper proposes a simple unsupervised multiplex graph diffusion network (UMGDN) that does not need to introduce any data augmentation, negative pairs sampling techniques, complex pretext tasks, and also adaptively aggregates the optimal frequency information between nodes to generate more robust node embeddings.
- 13
- Video scene parsing: An overview of deep learning methods and datasets13
This work describes several indispensable preliminaries of this field, defining essential background concepts as well as fundamental terminologies and differentiating between VSP and other similar problems, and draws conclusions on the state-of-the-art methods for VSP.
- 13
- Multiview partial least squares13
Experiments demonstrate that proposed multiview partial least squares is an effective and promising algorithm for practical applications.
- 13
- 13
- 13
- Rethinking attribute localization for zero-shot learning12
A novel attribute localization refinement (ALR) module is proposed designed to enhance the model’s ability to accurately localize all attributes by grouping them and introducing weighted attribute regression to standardize the mapping values of semantic attributes.
- SocialCircle+: Learning the Angle-based Conditioned Interaction Representation for Pedestrian Trajectory Prediction12
This work constructs an angle-based conditioned social interaction representation SocialCircle+ to represent the socially interactive context and its corresponding conditions and employs a social branch and a conditional branch to describe how pedestrians are positioned in prediction scenes socially and physically in angle-based-cyclic-sequence forms.
- 12
- Modal-Regression-Based Structured Low-Rank Matrix Recovery for Multiview Learning12
Different from previous LMvSL-based methods, this MR-SLMR can handle any zero-mode noise variable that contains a wide range of noise, such as Gaussian noise, random noise, and outliers and is robust to complicated noise.
- 12
- Characterization of Dirac Edge with New Wavelet Transform12
It is proven that the local maxima modulus of such continuous wavelet transform of a Dirac-structure edge forms two new curves which are located symmetrically at the two sides of the original one and have the same direction with it and the distance between the two curves is estimated.
- 11
- Computer-Aided Diagnosis of Congenital Abnormalities of the Kidney and Urinary Tract in Children Using a Multi-Instance Deep Learning Method Based on Ultrasound Imaging Data11
A robust pattern classifier was developed to distinguish children with CAKUT from controls based on their ultrasound images in sagittal and transverse views obtained during routine clinical care, built on imaging features derived using transfer learning from a pre-trained deep learning model with a mean pooling operator for fusing instance-level classification results.
- Texture Image Retrieval Using Novel Non-separable Filter Banks Based on Centrally Symmetric Matrices11
- A Unified B-Spline Framework for Scale-Invariant Keypoint Detection10
A unified B-spline framework for scale-invariant keypoint detection is proposed that provides a mathematical interpretation of existing fast detectors based on integral images and can be designed in a principled way without the heuristic choice of kernel shape and scales.
- IAS‐NET: Joint intraclassly adaptive GAN and segmentation network for unsupervised cross‐domain in neonatal brain MRI segmentation10
The proposed IAS-NET can improve the performance of the segmentation network effectively by its intra-class feature alignment in the target domain, and achieves state-of-the-art results in the compared UDA models for the segmentsation task.
- 10
- FINGERPRINT ENHANCEMENT BASED ON WAVELET AND ANISOTROPIC FILTERING10
A new method for fingerprint enhancement combining nontensor product wavelet filter banks and anisotropic filter is developed, capable of overcoming block effect and enhancing low quality fingerprint while preserving the ridge structures around singular points.
- 10
- Image matching using enclosed region detector10
A simple yet effective method to detect the affine-invariant regions from gray image, which is called enclosedregion, which outperforms state-of-the-art methods.
- FINGERPRINT ENHANCEMENT USING WAVELET TRANSFORM COMBINED WITH GABOR FILTER10
This paper combines the filter-based method with wavelet transform to achieve a more reliable and effective approach to fingerprint enhancement, which can distinctly improve the accuracy of the minutiae extraction module and finally achieve a better performance of the entire system.
- Detail Reinforcement Diffusion Model: Augmentation Fine-Grained Visual Categorization in Few-Shot Conditions9
The detail reinforcement diffusion model (DRDM) is proposed, which leverages the extensive knowledge of large models for fine-grained data augmentation and comprises two key components including discriminative semantic recombination (DSR) and spatial knowledge reference (SKR).
- 9
- MultiCut-MultiMix: a two-level data augmentation method for detecting small and densely distributed objects in large-size images9
A two-level data augmentation method, referred to as MultiCut-MultiMix, to solve the problems, in which MultiCut is developed to avoid feature distortion at the physical level and MultiMix is designed to enrich background at the pixel level for detecting small and densely distributed objects in large-size images.
- Generic Pixel Level Object Tracker Using Bi-Channel Fully Convolutional Network9
A novel bi-channel fully convolutional neural network is proposed to tackle the generic pixel-level object tracking problem and achieves compelling performance on challenging videos in comparison with competitive tracking algorithms.
- 9
- Self-adaptive local Fisher discriminant analysis for semi-supervised image recognition9
A self-adaptive semi-supervised dimension reduction framework for image class recognition and its performance is significantly influenced by some parameters - Neighbour size k of every labelled sample.
- 9
- 8
- Optimal Kernel Choice for Score Function-based Causal Discovery8
This paper model the generative process of the variables involved in each step of the causal graph search procedure as a mixture of independent noise variables, and derives an automatic kernel selection method by maximizing the marginal likelihood of the variables involved in each search step.
- Polylanenet++: enhancing the polynomial regression lane detection based on spatio-temporal fusion8
A polynomial regression-based deep learning model that enhances spatial generalization and incorporates temporal information to improve the accuracy and practicality in real-time applications is proposed.
- Towards Unsupervised Graph Completion Learning on Graphs with Features and Structure Missing8
A more general GCL framework with the aid of self-supervised learning for improving the task performance of the existing GNN variants on graphs with features and structure missing, termed unsupervised GCL (UGCL).
- 8
- Pedestrian Detection Based on Multi-Block Local Binary Pattern and Biologically Inspired Feature8
The MB-BIF approach combines two well-known methods, the Multi-Block Local Binary Pattern and Biologically Inspired Method, and shows good efficiency.
- MATCHING ALGORITHM USING WAVELET THINNING FEATURES FOR OFFLINE SIGNATURE VERIFICATION8
Experimental results showed that the proposed similarity measurement was able to provide sufficient discriminatory information in terms of equal error rate being 18.6% with four training samples.
- A Global-to-Local Matching Strategy for Registering Retinal Fundus Images8
A multi-resolution rigid-model-based global matching algorithm is employed to register tree structures of blood vessels extracted from retinal fundus images to improve alignment of the vessels and to eliminate the existence of ‘ghost vessels' for accurate registration.
- A Watermarking Scheme Based on Discrete Non-separable Wavelet Transform8
A blind watermarking scheme which is based on discrete non-separable wavelet transform (DNWT), which is more resistant to sharpening than DSWT scheme and more coefficients of DNWT can add watermark than that of discrete separable wavelets transform (DSWT).
- 7
- 7
- Adaptive Matching of Kernel Means7
A novel approach to adaptive matching of kernel means is proposed, and selected data with high importance are adopted to achieve calculation efficiency with optimization, and scalable learning can be conducted in proposed method as a generalized solution to matching of appended data.
- 7
- Generalized Kernel Normalized Mixed-Norm Algorithm: Analysis and Simulations7
Instead of using the original AMP proposed in the previous work, a novel time-varying AMP is employed to provide better tracking behavior to the nonstationarity and the mean square convergence analysis is conducted, where the second moment behavior of weight error vector is elaborately studied.
- A Batch Rival Penalized Expectation-Maximization Algorithm for Gaussian Mixture Clustering with Automatic Model Selection7
A batch Rival Penalized Expectation-Maximization (RPEM) algorithm for density mixture clustering provided that all observations are available before the learning process, which is faster than the EM and the adaptive RPEM in general.
- 7
- 7
- 7
- Region‐related focal loss for 3D brain tumor MRI segmentation6
The proposed Region-related Focal Loss (RFL) can converge with a invariant focalized point in the training of segmentation network, thus effectively alleviating the hard-voxel imbalance in brain tumor MRI segmentation.
- 6
- Self-Supervised Guided Hypergraph Feature Propagation for Semi-Supervised Classification with Missing Node Features6
A self-supervised guided hypergraph feature propagation (SGHFP), where the feature hypergraph is first generated according to the node features with missing information and then the reconstructed node features are fed to a two-layer GNNs to construct a pseudo-label hypergraph.
- Small infrared target detection based on image patch ordering6
The experimental results on two sequences show that the proposed patch ordering-based method can efficiently detect small IR target from heavy clutter.
- 6
- 6
- WAVELET NATURAL BOUNDARY ELEMENT METHOD FOR THE NEUMANN EXTERIOR PROBLEM OF STOKES EQUATIONS6
The Neumann exterior problem of Stokes equations is reduced into an equivalent Hadamard-singular Natural Integral Equation (NIE) by virtue of the wavelet-Galerkin algorithm, the simple and accurate computational formulae of stiffness matrix are obtained.
- MORPHOLOGICAL STRUCTURE RECONSTRUCTION OF RETINAL VESSELS IN FUNDUS IMAGES6
A two-stage scheme to extract vessels and reconstruct the morphological structure of vessels in retinal images using mathematical morphology techniques to highlight large and small vessels with respect to their spatial properties is proposed.
- 6
- 5
- Who Walks with You Matters: Perceiving Social Interactions with Groups for Pedestrian Trajectory Prediction5
The qualitative analysis indicates that the GPCC framework success-fully leverages grouping and perception cues human-like intuitively to validate the proposed model's explainability in pedestrian trajectory forecasting, and demonstrates significant improvements in trajectory prediction accuracy.
- 5
- Discriminative Image Generation with Diffusion Models for Zero-Shot Learning5
The extensive experiments and visualizations show that the DIG-ZSL generates diverse and high-quality images, outperforms previous state-of-the-art nonhuman-annotated semantic prototype-based methods by a large margin, and achieves comparable or better performance than baselines that leverage human-annotated semantic prototypes.
- Deep Supervised Information Bottleneck Hashing for Cross-modal Retrieval based Computer-aided Diagnosis5
This work proposes Deep Supervised Information Bottleneck Hashing (DSIBH), which effectively strengthens the discriminability of hash codes, and demonstrates the superior accuracy of the proposed DSIBH compared with state-of-the-arts in cross-modal medical data retrieval tasks.
- Group sparse additive machine with average top-k loss5
This paper employs the robust average top-k (ATk) loss as classification error measure and proposes a new sparse algorithm, named ATk group sparse additive machine ( ATk-GSAM), which has well adaptivity by integrating the data dependent hypothesis space and group sparse regularizer together.
- A joint space‐angle regularization approach for single 4D diffusion image super‐resolution5
A joint space‐angle regularization approach is proposed to reconstruct HSAR diffusion signals from a single 4D low resolution (LR) dMRI, which is down‐sampled in both 3D‐space and q‐space.
- Subsequent Boundary Distance Regression and Pixelwise Classification Networks for Automatic Kidney Segmentation in Ultrasound Images.5
This study proposes a novel data-augmentation method based on kidney shape registration to generate enriched training data from a small number of US images with manually segmented kidney labels, and demonstrates that this method could effectively improve the performance of automatic kidney segmentation, significantly better than deep learning based pixel classification networks.
- A Unified Approach for Spatial and Angular Super-Resolution of Diffusion Tensor MRI5
A new single image acquisition super-resolution method to increase both the spatial and angular resolution of dMRI, which combines the two strategies of nonlocal similarity reconstruction and compressed sensing reconstruction in a sparse basis of spherical ridgelets to reconstruct high resolution image in k-space with complex orientations.
- 5
- 5
- 5
- 4
- Mixed-norm partial least squares4
A new SPLS model — mixed-norm PLS (MNPLS) — to select common variables during each deflation is proposed in this paper and can generally outperform the standard PLS and other existing methods in variable selection and prediction.
- 4
- 4
- A New Approach for Spectra Baseline Correction using Sparse Representation4
Results show the highly efficiency and accuracy of the proposed algorithm for baseline correction for spectral signal based on sparse representation.
- 4
- 4
- Writer Identification Using a Hybrid Method Combining Gabor Wavelet and Mesh Fractal Dimension4
A new hybrid method combining Gabor function and mesh fractal dimension for off-line, text-independent writer identification is proposed, capable of extracting more useful features to distinguish the handwritings.
- 4
- Multiscale Approach for Thinning Ridges of Fingerprint4
The proposed multiscale thinning algorithm is able to improve the skeleton representation of the ridge of fingerprint without side-effects and limitations of previous methods.
- 3
- Generalized Sparse Additive Model with Unknown Link Function3
A new sparse additive model, named generalized sparse additive model with unknown link function (GSAMUL), is proposed, in which the component functions are estimated by B-spline basis and the unknown link function is estimated by a multi-layer perceptron (MLP) network.
- 3
- 3
- Person-Independent Emotion and Gender Prediction (EGP) System Using EEG Signals3
This paper presents a person-independent Emotion and Gender Prediction (EGP) system using Electroencephalography (EEG) brain signals and the highest average accuracy is obtained with CNN classifier in all emotion states including the state of mixing all emotions together.
- 3
- 3
- 3
- 3
- 3
- Locating Vessel Centerlines in Retinal Images Using Wavelet Transform: A Multilevel Approach3
A novel thinning method based on the wavelet transform with a multilevel scheme is presented and results show that the identified centerline of vascular trees are accurate by visual inspection and are useful for further applications.
- 2
- Leachable Component Clustering2
A novel approach to clustering of incomplete data, termed leachable component clustering, is proposed, and the proposed method handles data imputation with Bayes alignment, and collects the lost patterns in theory.
- 2
- Emotion Assessment Based on EEG Brain Signals2
This paper presents an emotion assessment method that classifies five emotions (happy, sad, angry, fear, and disgust) using EEG brain signals to produce robust system with high accuracy result.
- 2
- 2
- 2
- Efficient Group Learning with Hypergraph Partition in Multi-task Learning2
This paper describes the group learning process as permuting the parameter matrix into a block diagonal matrix, which can be modeled as a hypergraph partition problem, and demonstrates that the optimization algorithm scales well to large data.
- 2
- 2
- An Adaptive Image Watermarking Scheme Using Non-separable Wavelets and Support Vector Regression2
The superiority of non-separable wavelet transform (DNWT) in capturing image features combined with the good generalization ability of support vector regression provide us with a promising way to design a more robust watermarking algorithm featuring a better trade-off between the robustness and imperceptivity, the main duality of watermarked algorithms.
- 2
- 2
- A Novel Fusing Algorithm for Retinal Fundus Images2
A novel fusing method for fundus retinal images based on robust registration techniques is proposed and a ‘coarse-to-fine’ mapping strategy is applied to accurately align pairs of identified vascular trees of retinas.
- A WAVELET-BASED APPROACH TO RIDGE THINNING IN FINGERPRINT IMAGES2
A robust wavelet-based method to create thinned ridge map of fingerprint for automatic recognition is proposed and properties of modulus minima based on the spline wavelet function are substantially investigated.
- Similarity Measurement for Off-Line Signature Verification2
Experimental results showed that the computed similarity measurement was able to provide sufficient discriminatory information and performance in terms of equal error rate was 18.6% with four training samples.
- Skeletonization of Fingerprint Based-on Modulus Minima of Wavelet Transform2
Experimental results show that the skeletons of object obtained from the proposed algorithm overcome greatly some of the undesirable effects and limitations of previous methods, moreover, the proposed algorithms is insensitive to noise as well as efficient computability.
- Skeletonization of Character Based on Wavelet Transform2
A novel scheme for computing skeleton of characters using wavelet transform by using multiscale corner detection with new wavelet function, and a set of modifying techniques are developed to remove the artifacts of primary skeletons.
- Mutually Causal Semantic Distillation Network for Zero-Shot Learning1
A mutually causal semantic distillation network (termed MSDN++) is proposed to distill the intrinsic and sufficient semantic representations for ZSL to yield significant improvements over the strong baselines, leading to new state-of-the-art performances.
- 1
- 1
- Prototype-Guided Curriculum Learning for Zero-Shot Learning1
This work proposes a prototype-guided curriculum learning framework (dubbed as CLZSL), which mitigates instance-level mismatches through a Prototype-Guided Curriculum Learning (PCL) module and addresses class-level imprecision via a Prototype Update (PUP) module.
- 1
- Kernelized Similarity Learning and Embedding for Dynamic Texture Synthesis1
This article first put forward two hypotheses that are essential for the DT model to generate new frames using similarity correlations, and integrates kernel learning and the extreme learning machine into a unified synthesis model to learn kernel similarity embeddings for representing DTs.
- Information-Theoretic Hashing for Zero-Shot Cross-Modal Retrieval1
This paper considers a totally different way to construct (or learn) a common hamming space from an information-theoretic perspective and proposes the Information-Theoretic Hashing, which is composed of two cascading modules: an Adaptive Information Aggregation module; and a Semantic Preserving Encoding module.
- 1
- 1
- 1
- Similarity-DT: Kernel Similarity Embedding for Dynamic Texture Synthesis.1
This paper proposes a novel DT synthesis method, which embeds the similarity prior into the representation of DT, and integrates kernel learning and extreme learning machine (ELM) into a powerful unified synthesis model to learn kernel similarity embedding to represent the spatial-temporal transition among frame-to-frame of DTs.
- Retrieval by Classification: Discriminative Binary Embedding for Sketch-Based Image Retrieval1
This work proposes Discriminative Binary Embedding (DBE), a novel algorithm of considering inter-class relationship and object recognition ability in a joint manner by treating retrieval as classification and Experimental results on two benchmarks show that DBE outperforms several state-of-the-arts.
- Robust Multi-view Subspace Learning Through Structured Low-Rank Matrix Recovery1
This work gets rid of the framework of graph embedding and introduces class-label matrix to flexibly design a supervised low-rank model, which successfully learns a discriminative common subspace and discovers the invariant features shared by multi-view data.
- 1
- Robust Multi-view Common Component Learning1
This paper proposes robust multi-view common component learning (RMCCL) algorithm, which learns multiple linear transforms to extract the common component of multi- view data from the same instance, and introduces binary label matrix technology and Cauchy loss as the authors' error measurement.
- 1
- Learning a Sparse Representation for Robust Face Recognition1
The occlusions have sparse representation on the learned occlusion dictionary and can be effectively separated from the occluded face images, while the computational cost is much lower.
- 1
- 1
- Mosaicing the Retinal Fundus Images: A Robust Registration Technique Based Approach1
A simple rigid model is proposed to globally match vascular trees via a multi-resolution scheme and an elastic matching algorithm is employed to achieve accurate local alignment.
- 1
- –
- –
- –
- Efficient Invariant Representation Learning Based on Cauchy–Schwarz Divergence–
In CSIR, a novel method for learning invariant representations is proposed, termed Cauchy-Schwarz invariant representation learning (CSIR), which implements CMI and MI by leveraging the classic Cauchy-Schwarz divergence without the need for extra neural networks.
- –
- –
- –
- –
- WRF4CIR: Weight-Regularized Fine-Tuning Network for Composed Image Retrieval–
WRF4CIR, a Weight-Regularized Fine-tuning network for CIR, which increases the difficulty of fitting the training data, which helps mitigate overfitting in CIR under limited triplet supervision and achieves substantial improvements over existing methods.
- –
- –
- DH-MSVM: A hybrid algorithm for seeking quality support vectors in distributed learning–
This work proposes a Distributed Hybrid Learning based on Support Vector Machine (DH-SVM), which leverages global pre-learning to capture data structure information, which then guides local learning processes to identify higher-quality support vectors and adaptively refine decision boundaries.
- Robust training of open-set graph neural networks on graphs with in-distribution and out-of-distribution noise–
A novel RT-OGNN (robust training of open-set GNN) framework is proposed to weaken the impact of structure noise and extend the receptive field of nodes, which validate the superiority of RT-OGNN over state-of-the-art models.
- –
- –
- What Happens Without Background? Constructing Foreground-Only Data for Fine-Grained Tasks–
An engineered pipeline is proposed that leverages the capabilities of SAM and Detic to create fine-grained datasets with only foreground subjects, devoid of background, to facilitate research into the impact of background noise on models and enhance their ability to concentrate on the subject's discriminative features.
- FC2VR2: Few Critical Cues-aware Voice Relationship Representation for Speech Emotion Recognition with Transformer–
A novel few critical cues Transformer is proposed, which learns a few core features and long-distance features in speech and can achieve better performance than other state-of-the-art methods.
- –
- Deep Manifold Hashing: A Divide-and-Conquer Approach for Semi-Paired Unsupervised Cross-Modal Retrieval–
Deep Manifold Hashing is proposed, a novel method of dividing the problem of semi-paired unsupervised cross-modal retrieval into three sub-problems and building one simple yet efficiency model for each sub-problem.
- –
- Common Structured Low-Rank Matrix Recovery for Cross-View Classification–
This paper proposes a Common Structured Low-rank Matrix Recovery algorithm to elegantly handle view discrepancy and discriminancy simultaneously and incorporates common representation constraint and structured regularization into the fundamental model of LMvSL to learn a discriminant latent subspace.
- –
- –
- Saliency Detection by Compactness Diffusion–
This paper proposes saliency detecting with a diffusion model; use optimal seeds generated from foreground statistic cue, i.e., the compactness to calculate the global compactness of each node.
- –
- Webcam-Based Visual Gaze Estimation Under Desktop Environment–
This paper develops a low cost yet effective webcam based visual gaze estimation system that can achieve satisfactory performance without the requirements of dedicated hardware or tedious calibration procedure.
- Locally Linear Embedding Based Dynamic Texture Synthesis–
Experiments show that the proposed new framework for generating dynamic textures by using a new appearance model and a new observation model to preserves the non-linear correlation of video sequences is efficient in capturing complex appearance variation while maintaining the temporal coherence of the new synthesized texture videos.
- –
- –
- –
- –
- –
- Application of HHT method on feature extraction of GPR signal–
HHT method is adopted to analyze the feature of GPR signal and extract the instantaneous frequency of IMF components as feature vector and the result shows the technique of HHT is effective for G PR signal which can preferable avoid the influence of the directive wave.
- –
- –
- –
- –
- –
- –
- –
- LINEAR AND NONLINEAR FEATURE-BASED FUSION ALGORITHMS FOR FACE RECOGNITION–
This work proposes combining four linear and nonlinear face recognition algorithms via a weighted combination scheme to improve the recognition performance of a face recognition system and the results confirm the effectiveness of the proposed combination scheme.
- Axial Representation of Character by Using Wavelet Transform–
A novel scheme of extracting the central axis of character based on the wavelet transform (WT) is developed, applicable to both binary image and gray-level image and shows that the final axis obtained closely resembles the human perceptions.
- Skeleton Representation of Character Based on Multiscale Approach–
A novel algorithm based on multiscale approach to extract skeletons of printed and hand-written characters by thresholding for the modulus minima of wavelet transform, which can be exactly located on the central line of the stroke, and artifacts and branches from traditional methods can be avoided.
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.