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
- Hong YanSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works508 from public data
- Pyramid Dilated Deeper ConvLSTM for Video Salient Object Detection492
This paper proposes a fast video salient object detection model, based on a novel recurrent network architecture, named Pyramid Dilated Bidirectional ConvLSTM (PDB-ConvL STM), which achieves state-of-the-art results on two popular benchmarks, well demonstrating its superior performance and high applicability.
- 408
- Locating and extracting the eye in human face images381
The corner detection scheme introduced in this paper can provide accurate information about the corners and accurately locate the templates in relation to the eye images and greatly reduce the processing time for the templates.
- An analytic-to-holistic approach for face recognition based on a single frontal view284
An analytic-to-holistic approach which can identify faces at different perspective variations is proposed, and it is shown that this approach can achieve a similar level of performance from different viewing directions of a face.
- Underwater Camera: Improving Visual Perception Via Adaptive Dark Pixel Prior and Color Correction231
Experimental results show that the novel method for underwater image restoration outperforms state-of-the-art methods, demonstrating its efficacy in dealing with uneven lighting and diverse underwater environments.
- HCLR-Net: Hybrid Contrastive Learning Regularization with Locally Randomized Perturbation for Underwater Image Enhancement210
The HCLR-Net, a novel approach built upon a distinctive hybrid contrastive learning regularization strategy that incorporates a unique methodology for constructing negative samples, significantly enhances the visual quality of underwater images, establishing its innovative and practical applicability.
- An efficient illumination normalization method for face recognition179
An efficient representation method insensitive to varying illumination is proposed for human face recognition, which can effectively eliminate the effect of uneven illumination and greatly improve the recognition results.
- An efficient algorithm for human face detection and facial feature extraction under different conditions176
An efficient algorithm for human face detection and facial feature extraction is devised using the genetic algorithm and the eigenface technique, and the lighting effect and orientation of the faces are considered and solved.
- 170
- Face recognition under varying illumination based on a 2D face shape model169
A novel illumination compensation algorithm, which can compensate for the uneven illuminations on human faces and reconstruct face images in normal lighting conditions and eliminate the influence of uneven illumination while retaining the shape information about a human face is proposed.
- 151
- Extraction of the Euclidean skeleton based on a connectivity criterion134
A connectivity criterion is proposed, which can be used to determine whether a given pixel is a skeleton point independently, and is based on a set of point pairs along the object boundary, which are the nearest contour points to the pixel under consideration and its 8 neighbors.
- 129
- 129
- 128
- Person re-identification by unsupervised video matching120
This work proposes an unsupervised approach to person re-identification based on typical surveillance image-sequences, and presents a new video representation particularly tailored for person ReID, built up on existing action space-time features.
- Optimal sampling of Gabor features for face recognition117
Experimental results show that the new sampling method is simple, and effective for both dimension reduction and image representation, and the recognition rate based on the proposed scheme is higher than that achieved using a regular sampling method in a face region.
- 116
- 116
- 112
- Integrating QDWD with pattern distinctness and local contrast for underwater saliency detection109
Experimental results, based on the publicly available OUC-VISION underwater image database, show that the proposed method can produce reliable and promising results, compared to other state-of-the-art saliency-detection models.
- Using the K-Nearest Neighbor Algorithm for the Classification of Lymph Node Metastasis in Gastric Cancer100
GSI-CT can be the optimal choice for the preoperative diagnosis of patients with gastric cancer in the N staging, and the K-nearest neighbor classifier can be used to distinguish lymph node metastasis from nonlymph nodes metastasis.
- Generation of moment invariants and their uses for character recognition97
The generated sets of moment invariants, including higher-order moments, have been used for the recognition of English alphabets, such that the recognition rates have improved to over 98.8%.
- Illumination invariant face recognition96
A novel approach to handle the illumination problem that can restore a face image captured under arbitrary lighting conditions to one with frontal illumination by using a ratio-image between the face image and a reference face image, both of which are blurred by a Gaussian filter.
- Fast greedy algorithm for active contours96
The greedy algorithm is a fast iterative method for energy minimisation of snakes for image contour detection and an even faster algorithm is proposed, which reduces the computing time by 30% on average.
- Histogram-based local descriptors for facial expression recognition (FER): A comprehensive study91
Different studies of the visual features for FER are brought together by evaluating their performances under the same experimental setup, and critically reviewing various classifiers making use of the local descriptors.
- 91
- A robust scheme for live detection of human faces in color images89
Experimental results show that this algorithm can detect human faces under varying lighting conditions reliably and fast.
- 88
- 88
- Simplified Gabor wavelets for human face recognition88
Experimental results show that using SGWs can achieve a performance level similar to using GWs, while the runtime for feature extraction usingSGWs is, at most, 4.39 times faster than that of GWs implemented by using the fast Fourier transform (FFT).
- Multi-resolution feature fusion for face recognition78
A novel face recognition approach using information about face images at higher and lower resolutions so as to enhance the information content of the features that are extracted and combined at different resolutions and employs the cascaded generalized canonical correlation analysis (GCCA).
- Spatially eigen-weighted Hausdorff distances for human face recognition78
Two modified Hausdorff distances, namely, SEW2HD and SEW2HD are proposed, which incorporate the information about the location of important facial features such as eyes, mouth, and face contour so that distances at those regions will be emphasized.
- NormAttention-PSN: A High-frequency Region Enhanced Photometric Stereo Network with Normalized Attention77
An attention-weighted loss is presented to produce better surface reconstructions, which applies a higher weight to the detail-preserving gradient loss in high-frequency areas, and a double-gate normalization method for non-Lambertian surfaces is adopted.
- 75
- Deep Learning Methods for Calibrated Photometric Stereo and Beyond74
This paper provides a comprehensive review of existing deep learning-based calibrated photometric stereo methods utilizing orthographic cameras and directional light sources and summarizes the performance of deep learning photometric stereo models on the most widely-used benchmark data set.
- The extended marine underwater environment database and baseline evaluations72
An extended underwater image database for salient-object detection or saliency detection is introduced, called the Marine Underwater Environment Database (MUED), which contains 8600 underwater images of 430 individual groups of conspicuous objects with complex backgrounds, multiple salient objects, and complicated variations in pose, spatial location, illumination, turbidity of water, etc.
- Facial-feature detection and localization based on a hierarchical scheme72
An efficient hierarchical scheme, which is robust to illumination and pose variations in face images, is proposed for accurate facial-feature detection and localization and can achieve a superior performance compared to other state-of-the-art methods.
- 71
- An adaptive active contour model for highly irregular boundaries71
In this method, a contour can break automatically to represent the contours of multiple objects, and a better stopping criterion based on the area of a closed contour is devised.
- Facial expression recognition based on shape and texture69
A representation model for facial expressions, namely the spatially maximum occurrence model (SMOM), is proposed, which is based on the statistical characteristics of training facial images and has a powerful representation capability, and the elastic shape-texture matching (ESTM) algorithm is used to measure the similarity between images.
- FactLLaMA: Optimizing Instruction-Following Language Models with External Knowledge for Automated Fact-Checking68
This work proposes combining the power of instruction-following language models with external evidence retrieval to enhance fact-checking performance by leveraging search engines to retrieve relevant evidence for a given input claim and instruct-tune an open-sourced language model, called LLaMA, using this evidence.
- Deep-feature encoding-based discriminative model for age-invariant face recognition68
Experimental results show that the proposed robust deep-feature encoding-based discriminative model for age-invariant face recognition outperforms various state-of-the-art face recognition methods, in terms of the rank-1 recognition accuracy.
- Face hallucination based on sparse local-pixel structure68
The proposed face-hallucination method, namely face hallucination based on sparse local-pixel structure, aims to shape the prior model using sparse representation, and achieves a competitive or even superior performance in terms of both reconstruction error and visual quality.
- 64
- An efficient low bit-rate video-coding algorithm focusing on moving regions64
A very low bit-rate video coding algorithm that focuses on moving regions is proposed, which gives better subjective and objective quality than that of the conventional coding methods at the same bit rate.
- A fast fractal image coding based on kick-out and zero contrast conditions63
A fast algorithm for fractal image coding based on a single kick-out condition and the zero contrast prediction that can achieve the same reconstructed image quality as the exhaustive search, and can greatly reduce the required computation or runtime.
- 61
- 61
- Structured Adversarial Self-Supervised Learning for Robust Object Detection in Remote Sensing Images59
- 59
- Locating the eye in human face images using fractal dimensions59
A modified approach to estimating fractal dimensions that is less sensitive to lighting conditions and provides information about the orientation of an image under consideration is proposed.
- 58
- Example-based image super-resolution with class-specific predictors58
A novel learning-based algorithm for image super-resolution is presented to improve the computational speed and prediction accuracy and it is demonstrated that the new method provides improved performances over existing methods.
- A novel face-hallucination scheme based on singular value decomposition56
Compared to state-of-the-art algorithms, experiments show that the proposed face-hallucination scheme is effective in terms of producing plausible HR images with both a holistic structure and high-frequency details.
- Fast algorithm for locating head boundaries56
A reliable method to locate the approximate position of the head and to estimate the head boundary is proposed and a fast algorithm based on the greedy algorithm for active contour modeling is presented.
- PromoterExplorer: an effective promoter identification method based on the AdaBoost algorithm53
An effective promoter identification algorithm, which is called PromoterExplorer, is proposed, which analyzes the different roles of various features, that is, local distribution of pentamers, positional CpG island features and digitized DNA sequence, and combines them to build a high-dimensional input vector.
- Age-invariant face recognition based on identity inference from appearance age52
An age-invariant face recognition scheme based on appearance age based on age-subspace learning from appearance-age labels that can achieve a comparable, or even better, performance against other state-of-the-art methods, especially when the age range is large.
- Human face recognition based on spatially weighted Hausdorff distance52
A new modified Hansdorff distance measure is proposed, which has a better discriminant power and is weighted according to a weighted function derived from the spatial information of the human face; hence crucial regions are emphasized for face identification.
- Human Face Image Recognition: An Evidence Aggregation Approach52
A novel analytically-based face recognition system is presented which allows incorporation of the importance of individual facial components in the recognition task and a fuzzy information fusion technique is applied to combine the decisions of individual classifiers with all possible combinations of classifiers.
- 50
- 50
- 50
- 50
- 48
- 48
- 48
- 47
- 47
- 46
- Online Video Super-Resolution With Convolutional Kernel Bypass Grafts46
This paper proposes an extremely low-latency VSR algorithm based on a novel kernel knowledge transfer method, named the convolutional kernel bypass graft (CKBG), which can process online video sequences up to 110 FPS with very low model complexity and competitive SR performance.
- 46
- 44
- Illumination-insensitive texture discrimination based on illumination compensation and enhancement44
The advantage of the proposed method is that the assumption of a single-point light source is not required, so it circumvents and overcomes the limitations of the Lambertian model and is also suitable for outdoor settings.
- 43
- Frequency layered color indexing for content-based image retrieval43
A method which separates an image into layers, each of which retains only pixels in areas with similar spatial frequency characteristics and uses simple low-level features to index the layers individually is introduced.
- Face-image retrieval based on singular values and potential-field representation41
An efficient method based on singular values and potential-field representation is proposed for face-image retrieval and theoretically proves that the leading singular values of an image can be used as a rotation-shift-scale-invariant global feature.
- 40
- GR-PSN: Learning to Estimate Surface Normal and Reconstruct Photometric Stereo Images38
A novel method, namely GR-PSN, which learns surface normals from photometric stereo images and generates the photometric images under distant illumination from different lighting directions and surface materials and shows realistic rendering results on 100 different materials is proposed.
- Saliency detection using quaternionic distance based weber local descriptor and level priors38
Experimental results show that the proposed method is able to produce reliable and promising salient maps/estimations, compared to other state-of-the-art saliency-detection models.
- Attention-ConvNet Network for Ocean-Front Prediction via Remote Sensing SST Images35
This study proposes a temporal-sensitive network named Attention-ConvNet, which significantly outperforms conventional methods, confirming its capability for precise ocean-front prediction.
- AGTGAN: Unpaired Image Translation for Photographic Ancient Character Generation34
This paper proposes an unsupervised generative adversarial network called AGTGAN, which can generate characters with diverse glyphs and realistic textures and outperforms other state-of-the-art methods in terms of various metrics and performs much better interms of the diversity and authenticity of generated samples.
- Content-based image retrieval via a hierarchical-local-feature extraction scheme34
A novel hierarchical-local-feature extraction scheme for CBIR, whereas complex image segmentation is avoided, and experimental results show that the developed CBIR system produces plausible retrieval results.
- Face Recognition with Multi-Resolution Spectral Feature Images34
A more accurate spectral feature image-based 2DLDA (two-dimensional linear discriminant analysis) ensemble algorithm for face recognition, with one sample image per person, inspired by the finding that, among these spectral feature images, features extracted from some orientations and scales using 2 DLDA are not sensitive to variations of illumination and expression.
- Recovering the 3D shape and poses of face images based on the similarity transform34
Experimental results show that the proposed algorithm can construct a 3D face structure reliably and efficiently and can be extended to face recognition to alleviate the effect of pose variations.
- 33
- 32
- An adaptive algorithm for the display of high-dynamic range images32
A novel algorithm based on spatial and statistical information is proposed for the display of high-dynamic range (HDR) images and an adaptive detail-enhancement method is proposed to combine the local and global issues.
- 31
- 31
- Computer-Aided Diagnosis for Distinguishing Pancreatic Mucinous Cystic Neoplasms From Serous Oligocystic Adenomas in Spectral CT Images30
The combination of conventional features and additional information provided by dual-energy spectral CT shows a high accuracy in the proposed computer-aided diagnosis (CAD) scheme and may prove useful in the diagnosis and classification of SOAs and MCNs with machine learning algorithms.
- 30
- 30
- 29
- Improved searching scheme for fractal image coding29
Experimental results show that the proposed scheme can achieve a speedup of 50% with only slight degradation in PSNR and compression ratio.
- 28
- 28
- 28
- Revisiting One-Stage Deep Uncalibrated Photometric Stereo via Fourier Embedding27
A one-stage deep uncalibrated photometric stereo (UPS) network, namely Fourier Uncalibrated Photometric Stereo Network (FUPS-Net), is introduced, offering better training stability, a concise end-to-end structure, and avoiding accumulated errors in disjointed networks.
- 27
- Deep Multi-task Learning for Facial Expression Recognition and Synthesis Based on Selective Feature Sharing27
A novel selective feature-sharing method is proposed, and a multi-task network for facial expression recognition and facial expression synthesis is established, which can effectively transfer beneficial features between different tasks, while filtering out useless and harmful information.
- 27
- Face detection using simplified Gabor features and hierarchical regions in a cascade of classifiers27
It is argued that the context information about a face can be advantageously used in the deeper nodes of the boosting process when the features derived from the central region of the face do not provide any further benefit.
- Progressive and Selective Fusion Network for High Dynamic Range Imaging26
This paper proposes a novel method that can better fuse the features based on two ideas; one is multi-step feature fusion; the network gradually fuses the features in a stack of blocks having the same structure.
- Face recognition using elastic local reconstruction based on a single face image26
A new face recognition algorithm based on a single frontal-view image for each face subject, which considers the effect of the face manifold structure is proposed, which can greatly improve the recognition rates under all the different conditions.
- 25
- 24
- 23
- Saliency detection based on singular value decomposition23
Experimental results prove that the methods based on SVD achieve superior performance compared to other state-of-the-art methods for human-eye fixations, as well as salient-object detection, in terms of the area under the receiver operating characteristic (ROC) curve (AUC) score, the linear correlation coefficient (CC), the normalized scan-path saliency (NSS) score
- 23
- Elastic shape-texture matching for human face recognition23
A novel, elastic, shape-texture matching method, namely ESTM, for human face recognition is proposed, which can always achieve a better performance than other face recognition algorithms under different conditions, except when an image is under poor and uneven illumination.
- 23
- 22
- 22
- 22
- An efficient two-stage framework for image annotation22
This paper introduces a simple label-filtering algorithm, and presents a two-stage method for multi-class image labeling that prunes specific features for each single label and formalizes the annotation task as a discriminative classification problem.
- 21
- 20
- 20
- 20
- Self-feature Learning20
A dynamic, channel-agnostic filtering method that not only adaptively generates convolutional kernels based on the local information of each position, but also can significantly reduce the cost of computing the inter-channel redundancy.
- Computer-Aided Diagnosis for Preoperative Invasion Depth of Gastric Cancer with Dual-Energy Spectral CT Imaging20
The CAD system designed using machine-learning algorithms may be used to improve the identification accuracy in the assessment of serosal invasion in patients of gastric cancer with DEsCT imaging and provide some indicators which may be useful in predicting prognosis.
- See In Detail: Enhancing Sparse-view 3D Gaussian Splatting with Local Depth and Semantic Regularization19
This work proposes a semantic regularization technique, using features extracted from the pretrained DINO-ViT model, to ensure multi-view semantic consistency and proposes local depth regularization, which constrains depth values to improve generalization on unseen views.
- 19
- Image super-resolution via feature-augmented random forest19
A novel feature-augmented random forest (FARF) for image super-resolution, where the conventional gradient-based features are augmented with gradient magnitudes and different feature recipes are formulated on different stages in an RF.
- 19
- 19
- 18
- 18
- 18
- Saliency detection based on background seeds by object proposals and extended random walk18
A novel bottom-up saliency-detection algorithm is proposed to tackle and overcome the issue of undesired errors and retrieval outputs when the boundaries of the salient objects concerned touch, or connect with, the image’s border.
- Efficient saliency analysis based on wavelet transform and entropy theory18
Experimental results show that the proposed saliency-detection method, based on wavelet transformation and entropy theory, can achieve excellent performance in terms of the area under the receiver operating characteristic curve (AUC), the linear correlation coefficient (CC), the normalized scan-path saliency (NSS) score, and visual performance, as compared to existing state-of-the-art methods.
- Adaptive deformable model for mouth boundary detection18
The results indicate that the new model adaptation technique sat-isfactorily generalizes the mouth boundary model extraction in an auto-mated fashion.
- STFE-Net: A Spatial-Temporal Feature Extraction Network for Continuous Sign Language Translation17
A spatial-temporal feature extraction network (STFE-Net) is proposed for CSLT, which optimally fuses spatial and temporal features, extracted by the spatial feature extractionNetwork (SFE- net) and the temporal feature extractionnetwork (TFE-Net), respectively.
- 17
- 17
- 17
- Design and learn distinctive features from pore-scale facial keypoints17
The results prove that the pore-scale features are highly distinctive; face images with a minimum resolution of 600i?700 (0.4 mega) pixels contain sufficient details to perform a reliable matching in different poses.
- 17
- 17
- Visual saliency detection via combining center prior and U-Net16
A traditional center prior-based method and U-Net model were combined efficiently to improve background suppression results and has been tested with four widely used databases and proven effective by comparing its results with those of another seven popular methods.
- 16
- 15
- Balanced distortion and perception in single-image super-resolution based on optimal transport in wavelet domain15
Compared with existing state-of-the-art SISR methods, experiment results show that the proposed image-fusion interpolation method can achieve a better balance between the distortion and visual quality in super-resolved images.
- 15
- 15
- Image magnification based on a blockwise adaptive Markov random field model15
An effective image magnification algorithm based on an adaptive Markov random field model with a Bayesian framework can provide a magnified image with the well-preserved edges and texture, and can achieve a better PSNR and visual quality.
- Efficient rotation- and scale-invariant texture classification method based on Gabor wavelets15
An efficient texture classification method is proposed that considers the effects of both the rotation and scale of texture and an elastic cross- frequency searching mechanism is devised to reduce the effect of scaling.
- Multi-Factor Deep Learning Model for Sea Surface Temperature Forecasting14
A multi-sensor SST prediction model that integrates Long Short-Term Memory networks, convolutional neural networks, and an attention mechanism to directly incorporate physical variables such as temperature, salinity, density, and current velocity is proposed.
- Point Cloud Densification for 3D Gaussian Splatting from Sparse Input Views14
This work proposes a densification method that generates high-quality point clouds, providing a superior initialization for 3D Gaussians, and proposes Systematically Angle of View Sampling (SAOVS), which employs Spherical Linear Interpolation and linear interpolation for side view sampling, to determine unseen views outside the training data for semantic pseudo-label regularization.
- 14
- 14
- 14
- MGF6mARice: prediction of DNA N6-methyladenine sites in rice by exploiting molecular graph feature and residual block14
A novel deep learning method is proposed by devising DNA molecular graph feature and residual block structure for 6mA sites prediction in rice, named MGF6mARice, which is the first time to derive a feature of DNA sequence by considering the chemical molecular structure.
- 14
- Shape-appearance-correlated active appearance model14
A shape-appearance-correlated AAM is proposed to tackle the generic problem of localizing feature points on unseen faces, and the optimization is solved by using the recently proposed fast simultaneous inverse compositional (Fast-SIC) algorithm.
- 14
- 14
- Eigentransformation-based face super-resolution in the wavelet domain14
By combining interpolation and eigentransformation, the reconstructed images are less dependent on the training set selected, and can better preserve the low-frequency content.
- A novel kernel-based framework for facial-image hallucination14
Experimental results show that the proposed kernel-based eigentransformation framework, along with a proper kernel, can produce good HR facial images in terms of both visual quality and reconstruction errors.
- 14
- 14
- 14
- An Efficient Algorithm for Ocean-Front Evolution Trend Recognition13
A novel method, named Evolution Trend Recognition (ETR), is proposed to recognize the trend of ocean fronts, being the most important information in the ocean dynamic process, and results show that the proposed ETR algorithm is highly promising for trend classification.
- 13
- 13
- 13
- Video-object segmentation and 3D-trajectory estimation for monocular video sequences13
A video-object segmentation and 3D-trajectory estimation method for the analysis of dynamic scenes from a monocular uncalibrated view that can be employed for video-Object segmentation, 2D-to-3D video conversion, video- object retrieval, etc.
- 13
- Geometric Distortion Guided Transformer for Omnidirectional Image Super-Resolution12
This paper introduces a novel Geometric Distortion Guided Transformer for Omnidirectional image Super-Resolution (GDGT-OSR), and presents extensive experimental results on public datasets and shows that the new GDGT-OSR outperforms methods in existing literature.
- Fault Diagnosis of Rolling Bearings Using Denoising Multi-Channel Mixture of CNN and Mamba-Enhanced Adaptive Self-Attention LSTM12
The Mamba-enhanced adaptive self-attention long short-term memory (MASA-LSTM) model, which effectively captures both global and local dependencies in ultra-long time series data and addresses the limitations of traditional models in extracting long-range dependencies from such signals.
- 12
- 12
- Residual Attention Single-Head Vision Transformer Network for Rolling Bearing Fault Diagnosis in Noisy Environments12
The results demonstrate that the RA-SHViT-Net outperforms state-of-the-art methods in terms of accuracy and robustness, particularly in scenarios involving complex and noisy environments.
- 12
- See Clearly in the Distance: Representation Learning GAN for Low Resolution Object Recognition12
A Representation Learning Generative Adversarial Network (RL-GAN) to generate super image representation that is optimized for recognition, which improves the classification results significantly, with 10–15% gain on average, compared with benchmark solutions.
- Towards Accurate Pulmonary Nodule Detection by Representing Nodules as Points With High-Resolution Network12
This paper presents a novel representation for detecting nodules, in terms of their 3D center locations, which reduces the number of hyper-parameters and the corresponding computation related to anchors, thus making the nodule detection pipeline much simpler.
- A novel correspondence-based face-hallucination method12
The proposed technique first solves for appearance of the HR faces form the correspondence derived between an interpolated LR face and its corresponding HR face, and incorporates optical flow to solve the local structure at sub-pixel level for theHR faces.
- 12
- 12
- 12
- 11
- 11
- 11
- 11
- Subspace learning for facial expression recognition: an overview and a new perspective11
Experiments show that SLPM effectively reduces the dimensionality of the feature vectors and enhances the discriminative power of the extracted features, and experimental results also demonstrate that the learned deep features regularized by SLP acquire a better discriminability and generalizability for facial expression recognition.
- 11
- Learning sparse discriminant low-rank features for low-resolution face recognition11
The approach first decomposes a multiple of extracted local features into a set of representative basis and sparse error matrix, and then learns a projection matrix based on the proposed sparse-coding-based algorithm, which preserves the sparse structure of the learned low-rank features, in a low-dimensional feature subspace.
- 11
- Enhancement of a CNN-Based Denoiser Based on Spatial and Spectral Analysis11
A discrete wavelet denoising CNN (WDnCNN), which restores images corrupted by various noise with a single model, and employs a band normalization module (BNM) to normalize the coefficients from different parts of the frequency spectrum.
- 11
- FSAM: A fast self-adaptive method for correcting non-uniform illumination for 3D reconstruction11
Average absolute gradient (AAG) is proposed to compare the surface heightmaps reconstructed using corrected illumination with those reconstructed without illumination correction, and shows that the method is efficient and can produce convincing results.
- Visual sensitivity-based low-bit-rate image compression algorithm11
The authors algorithm combines both visual sensitivity and compression techniques so that a higher compression rate, with satisfactory visual quality, can be achieved.
- 11
- Complexity scalable control for H.264 motion estimation and mode decision under energy constraints11
The proposed algorithm can adaptively determine an appropriate cut-off point for a sequence of re-ordered candidate modes according to the current energy condition of a mobile device, which can adjust the complexity of inter-prediction coding to any appropriate level with minimum degradation in video quality and prolong the operational lifetime of batteries.
- 11
- 11
- 10
- 10
- 10
- Spectrally Layered Color Indexing10
This paper introduces a method, which separates an image into layers, each of which retains only pixels in areas with similar spectral distribution characteristics, thereby greatly enhancing the power of color indexing while keeping its simplicity and elegance.
- 10
- 9
- 9
- 9
- 9
- 9
- 9
- 9
- 9
- 9
- Facial Feature Location and Extraction for Computerized Human Face Recognition9
The head boundary is first located in a head-and-shoulders image and Corners, the salient features of the eyes, are detected and used to set the initial parameters of the eye templates.
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- Efficient color face detection algorithm under different lighting conditions8
In this algorithm, skin-colored pixels are identified using a region-based approach, which can provide more reliable skin color segmentation under various lighting conditions and the face boundary region of a face is verified by a probabilistic approach to reduce the chance of false alarms.
- 8
- 7
- 7
- Mode Information Guided CNN for Quality Enhancement of Screen Content Coding7
This work proposes a Mode Information guided CNN (MICNN) to further improve the quality of screen content sequences at the decoder side, and extracts the mode information from the bitstream as the input of MICNN.
- Two-dimensional multi-scale perceptive context for scene text recognition7
A new Two-Dimensional Multi-Scale Perceptive Context (TDMSPC) module is proposed, which performs multi-scale context learning, along both the horizontal and vertical directions, and then merges them, which can generate shape and layout-dependent feature maps for scene text recognition.
- NTGAN: Learning Blind Image Denoising without Clean Reference7
A novel deep unsupervised image-denoising method is proposed by regarding the noise reduction task as a special case of the noise transference task, which enables the network to acquire the denoising ability by only observing the corrupted samples.
- 7
- 7
- 7
- 7
- 7
- 7
- TRECVID2005 Experiments in The Hong Kong Polytechnic University: Shot Boundary Detection Based on a Multi-Step Comparison Scheme.7
By measuring the difference between frames at varying distances apart, a distance map is generated and used to determine the existence of a transition and its type and based on the distance map, the different transitions can be detected and classified easily.
- 7
- 7
- Point Clouds are Specialized Images: A Knowledge Transfer Approach for 3D Understanding6
PCExpert is presented, a novel SSRL approach that reinterprets point clouds as “specialized images”, and allows PCExpert to leverage knowledge derived from large-scale image modality in a more direct and deeper manner, via extensively sharing the parameters with a pre-trained image encoder in a multi-way Transformer architecture.
- 6
- Attention-based cross-modality interaction for multispectral pedestrian detection6
An attention-based cross-modality interaction (ACI) module is proposed, which aims to adaptively highlight and aggregate the discriminative regions and channels of the feature maps from RGB and thermal images, for illumination-invariant pedestrian detection.
- 6
- Super-resolution on remote sensing images6
A novel model architecture based on the very deep super-resolution (VDSR) model is proposed, which performs better than the state-of-the-art models on SST frames and evaluates the peak signal-to-noise ratio (PSNR) and perceptual loss of the model trained on the natural images and S ST frames.
- Joint maximum purity forest with application to image super-resolution6
Experiments showed that JMPF remarkably outperforms other state-of-the-art random-forest-based approaches for classification, clustering, and regression tasks, and was applied to image super-resolution specifically because the transformed, compact features are more discriminative to the clustering-regression scheme.
- 6
- Cascaded face alignment via intimacy definition feature6
This work proposes a random-forest based, cascaded regression model for face alignment by using a locally lightweight feature, namely intimacy definition feature, which is more discriminative than the pose-indexed feature, more efficient than the histogram of oriented gradients feature and the scale-invariant feature transform feature, and more compact than the local binary feature (LBF).
- Pore-Scale Facial Features Matching Under 3D Morphable Model Constraint6
This work adopts Pore Scale-Invariant Feature Transform (PSIFT) to extract pore-scale facial features from face images, and uses 3D Dense Face Alignment (3DDFA) to obtain a fitted 3D morphable model, which is constrained by matching keypoints.
- 6
- 6
- 6
- Image Magnification Based on Fractal Codes and Model Constraint6
An algorithm based on fractal codes and model constraint is proposed and shows that the proposed approach performs better than the other fractal based magnification method.
- 6
- Computing the inverse DFT with the in-place, in-order prime factor FFT algorithm6
A method for computing the inverse discrete Fourier transform (IDFT) by the in-place, in-order prime factor FFT algorithm (PFA) by modifying the input and the output index mapping equations.
- 5
- 5
- Multi-kernel feature extraction with dynamic fusion and downsampled residual feature embedding for predicting rice RNA N6-methyladenine sites5
A new end-to-end deep learning framework is developed, MFDm, to address issues faced on rice, including sparse information and invalid padding, reducing feature extraction accuracy.
- Spatio-temporal feature learning for enhancing video quality based on screen content characteristics5
- Dual autoencoder based zero shot learning in special domain5
A dual autoencoder model is proposed, which contains two similar encoder–decoder structures, to tackle the zero-shot recognition problem and achieves competitive performance compared with six recent state-of-the-art methods.
- 5
- 5
- 5
- An Effective Membership Probability Representation for Point Set Registration5
A new representation is proposed for membership probability of Gaussian mixture model, by utilizing two types of feature descriptor, i.e. shape context or fast point feature histograms, which is more robust to deformation, outlier, occlusion, and rotation.
- Self-supervised depth completion with attention-based loss5
A self-supervised framework that can generate accurate dense depth from RGB images and sparse depth without the need for dense depth labels is proposed and a novel attention-based loss that takes into account the statistical properties of the depth value distribution is proposed.
- 5
- 5
- 5
- Fast Eye Detection and Localization Using a Salient Map5
The proposed algorithm is robust to illumination variations, size, and orientation of face images, and uses a wavelet-based salient map, which can detect and reflect the most visually meaningful regions for eye detection and localization.
- 5
- Wavelet-Based Eigentransformation for Face Super-Resolution5
In this paper, a new approach to human face hallucination based on eigentransformation is proposed, where a face image is decomposed into different frequency bands using wavelet transform, so that different approaches can be applied to the low-frequency and high-frequency contents for increasing the resolution.
- Canny Edge Detection Using Bilateral Filter on Real Hexagonal Structure5
This paper presents a Canny edge detection method based on bilateral filtering which achieves better performance than single Gaussian filtering and is achieved on sampled images represented on a real hexagonal structure.
- 5
- 5
- 5
- 5
- 5
- 5
- 5
- 5
- 5
- Safeguarding Vision-Language Models: Mitigating Vulnerabilities to Gaussian Noise in Perturbation-Based Attacks4
Robust-VLGuard is proposed, a multimodal safety dataset with aligned / misaligned imagetext pairs, combined with noise-augmented fine-tuning that reduces attack success rates while preserving functionality of VLM, and DiffPure-VLM, leveraging diffusion models to convert adversarial perturbations into Gaussian-like noise, which can be defended by VLMs with noise-augmented safety fine-tuning.
- SA2Net: Scale-adaptive structure-affinity transformation for spine segmentation from ultrasound volume projection imaging4
The proposed novel scale-adaptive structure-aware network (SA2Net) achieves superior segmentation performance compared to other state-of-the-art methods and the adaptability of SA2Net to various backbones enhances its potential as a promising tool for advanced scoliosis diagnosis using intelligent spinal image analysis.
- 4
- 4
- 4
- Restoration of User Videos Shared on Social Media4
This paper presents a new general video restoration framework that introduces the concept of a degradation feature map (DFM) to explicitly guide the video restoration process and presents extensive experimental results to show that the new VOTES method outperforms various state-of-the-art techniques both quantitatively and qualitatively.
- 4
- Robust seed selection of foreground and background priors based on directional blocks for saliency-detection system4
This paper explores the local geometry and statistical attribute of the detected orientational blocks via an improved discrete wavelet frame transform algorithm to estimate the center position of individual salient object in the original input.
- Deep Motion-Appearance Convolutions for Robust Visual Tracking4
A deep neural network for visual tracking, namely the Motion-Appearance Dual (MADual) network, which employs a dual-branch architecture by using deep two-dimensional and deep three-dimensional convolutions to integrate the local and global information of the target object’s motion and appearance synchronously.
- Face hallucination using orthogonal canonical correlation analysis4
Compared with other state-of-the-art approaches, the proposed framework can achieve a comparable, or even better, performance in terms of global face reconstruction and the visual quality of face hallucination.
- 4
- 4
- A Novel Low-Bit-Rate Image Compression Algorithm4
A novel image compression approach for low-bit-rate applications that combines both super-resolution techniques and compression techniques so that a higher compression rate, with satisfactory visual quality, can be achieved.
- 4
- 4
- 4
- 4
- HFS: Holistic Query-Aware Frame Selection for Efficient Video Understanding3
This work proposes an end-to-end trainable, task-adaptive framework for frame selection that significantly outperforms existing approaches, and defines a continuous set-level objective function that incorporates relevance, coverage, and redundancy.
- 3
- 3
- 3
- Enhancing Technical Documents Retrieval for RAG3
The approach addresses the challenges of understanding and retrieving complex technical content by leveraging the capabilities of Large Language Models by generating expanded representations that better capture user intent and improve dataset diversity, thereby enriching the fine-tuning process for embedding models.
- Semi-supervised learning for compound facial expression recognition with basic expression data3
This work proposes a semi-supervised learning framework to generate pseudo-compound-emotion labels for basic expression data and explores leveraging the basic labels by introducing a basic-label smoothing mechanism.
- 3
- 3
- Deep Progressive Feature Aggregation Network for High Dynamic Range Imaging3
Experiments show that the proposed method can achieve state-of-the-art performances under diverse scenes, compared to other promising HDR imaging methods, and the HDR images generated by the method contain cleaner and more detailed content, with fewer distortions, leading to better visual quality.
- 3
- Constructing an efficient and adaptive learning model for 3D object generation3
This work proposes a 3D spatial transformer network, and integrates it into a generation model, whose ability for extracting and reconstructing features for 3D objects can be improved, and develops a point-wise approach to reduce the computational complexity of the proposed network.
- 3
- 3
- 3
- 3
- 3
- Gabor-Feature Hallucination based on Generalized Canonical Correlation Analysis for face recognition3
- Learning a discriminative model for image annotation3
A new discriminative model for image annotation is introduced that divides each training image into patches, and embeds the patches into a hypergraph, so as to find the representative instances for every single class by solving the graph.
- 3
- Illumination Compensation3
This paper presents an illumination compensation scheme, whereby the uneven illumination of human faces is compensated and face images in normal lighting conditions are reconstructed and then used for classification.
- An image magnification algorithm using the GVF constraint model3
Experimental results show that the proposed method can improve the quality of magnified images in terms of both objective and subjective criteria.
- 3
- 3
- Semi-supervised Learning based on Bayesian Networks and Optimization for Interactive Image Retrieval3
This work shows that this semi-supervised learning method can be naturally adopted as a computational tool to incorporate users feedbacks for interactive image retrieval and shows that semi- supervised learning can have advantages over supervised and unsupervisedLearning in image retrieval applications.
- Image Synthesis and Face Recognition Based on 3D Face Model and Illumination Model3
A novel method that can synthesize images with different head poses and lighting conditions by using a modified 3D CANDIDE model, linear vertex interpolation and NURBS curve surface fitting method, as well as a mixed illumination model is proposed.
- An efficient video shot representation for fast video retrieval3
This paper proposes an efficient representation scheme for a shot, which considers both the spatial frequency contents and the temporal statistics of the frames for video retrieval, and outperforms the alpha-trimmed average histogram method in terms of retrieval accuracy.
- 3
- 3
- A Robust Line-Feature-Based Hausdorff Distance for Shape Matching3
A robust line-feature-based approach for model-based recognition is proposed, which can achieve a good performance level in matching, even in a noisy environment or with the existence of occlusion.
- Automatic Human Face Recognition System Using Fractal Dimension and Modified Hausdorff Distance3
A modified approach to estimate the fractal dimensions which is less sensitive to lighting conditions and provides information about the orientation of an image under consideration and can achieve recognition rates of 76%, 84%, and 92% for the first one, the first five, first ten likely matched faces, respectively.
- 2
- 2
- Adapting Large VLMs with Iterative and Manual Instructions for Generative Low-light Enhancement2
This paper proposes VLM-IMI, a framework that adapts large vision-language models with iterative and manual instructions for generative LLIE and introduces a learnable instruction prior fusion module, which dynamically aligns and fuses image and text features, promoting the generation of detailed and semantically coherent outputs.
- HAAT: hybrid attention aggregation transformer for image super-resolution2
A novel model, the Hybrid Attention Aggregation Transformer (HAAT), designed to better leverage feature information and surpasses state-of-the-art methods on benchmark datasets is introduced.
- AFSRNet: learning local descriptors with adaptive multi-scale feature fusion and symmetric regularization2
This paper proposes a novel convolutional neural network, based on center-surround adaptive multi-scale feature fusion, and introduces a novel regularization technique that uses second-order similarity to constrain the learning of local descriptors, based on the symmetric property of the similarity matrix.
- 2
- 2
- An End-to-End Two-Stream Network Based on RGB Flow and Representation Flow for Human Action Recognition2
This work introduces a representation flow algorithm and replaces the optical-flow branch in the egocentric action recognition model with this new branch for end-to-end training, which can greatly reduce computational cost and prediction runtime.
- Restoration of Multiple Image Distortions using a Semi-dynamic Deep Neural Network2
This paper presents a dynamic universal image restoration (DUIR) system capable of simultaneously processing multiple distortions, and shows that the new technique achieves superior performance to state-of-the-art models on both synthetic and real data.
- Unifying Multimodal Source and Propagation Graph for Rumour Detection on Social Media with Missing Features2
Experimental results show that the proposed Transformer-Graph Neural Network can achieve state-of-the-art performance and generalization ability evaluated on the PHEME-2016, PHEME-2018, and Weibo data sets.
- Multi-level unsupervised domain adaption for privacy-protected in-bed pose estimation2
This paper solves a crossdomain pose estimation problem, in which a fully annotated uncovered training set is used for pose estimation learning, and a large-scale unlabelled data set of covered images is employed for unsupervised domain adaptation.
- Application of GoogLeNet for ocean-front tracking2
This paper proposes an ocean-front tracking dataset (OFTraD) and applies GoogLeNet inception network to track ocean fronts in video sequences and shows that the algorithm can achieve accurate tracking results.
- 2
- Cascaded one-vs-rest detection network for fine-grained recognition without part annotations2
This paper proposes a novel cascaded deep CNN detection framework for fine-grained recognition which is trained to detect a whole object without considering parts, and introduces a novel one-vs-rest loss function to capture more minute variances from different subordinate categories.
- 2
- 2
- Efficient likelihood Bayesian constrained local model2
This paper presents a novel CLM variant, which employs random-forest regressors to estimate the location of each landmark, as a likelihood term, efficiently, and is called efficient likelihood Bayesian constrained local model (elBCLM).
- 2
- Color correction with blind image restoration based on multiple images using a low-rank model2
A method that can handle the color correction of multiple photographs with blind image restoration simultaneously and automatically is presented and it is proved that the local colors of a set of images of the same scene exhibit the low-rank property locally both before and after a color-correction operation.
- 2
- 2
- Pseudo-Gabor wavelet for face recognition2
An efficient face-recognition algorithm is proposed, which not only possesses the advantages of linear subspace analysis approaches—such as low computational complexity—but also has the advantage of a high recognition performance with the wavelet-based algorithms.
- 2
- 2
- 2
- 2
- 2
- 2
- AN EFFECTIVE PROMOTER DETECTION METHOD USING THE ADABOOST ALGORITHM2
The proposed promoter detection algorithm consistently outperforms PromoterInspector and Dragon Promoter Finder and is tested based on large-scale DNA sequences from different databases, including EPD, Genbank and human chromosome 22.
- Region of interest based image coding and progressive transmission2
An arbitrary shape integer or non-integer wavelet transform is employed to implement the lossless compression of region of interest (ROI) with arbitrary shape and lossy compression of non-ROI with highly efficient SPIHT algorithm.
- Robust Hausdorff distance for shape matching2
A robust line-feature-based approach for model-based recognition is presented, which can achieve good matching, even in a noisy environment or with the existence of occlusion, and can greatly reduce the required memory and computation when compared to a 4D-matching.
- 1
- 1
- 1
- 1
- 1
- 1
- Triple branch deep network for polyp image segmentation1
A triple-branch medical image segmentation network (TBIB-Net) that incorporates implicit boundary priors and the boundary map, acquired by a boundary detection algorithm, is used to restrict the results of the boundary branch.
- MFGAN: OCT image super-resolution and enhancement with blind degradation and multi-frame fusion1
This paper proposes a generative adversarial network (GAN) based OCT image super-resolution framework that leverages a blind degradation and Multi-frame Fusion mechanism, namely MFGAN, for retinal OCT image super-resolution.
- Enhancing Novel View Synthesis from extremely sparse views with SfM-free 3D Gaussian Splatting Framework1
A novel SfM-free 3DGS-based method that jointly estimates camera poses and reconstructs 3D scenes from extremely sparse-view inputs is proposed, and multi-scale Laplacian consistent regularization and adaptive spatial-aware multi-scale geometry regularization are introduced to enhance the quality of geometrical structures and rendered content.
- Spatial-frequency fusion for arbitrary-scale ultra-high-definition image super-resolution1
This work employs an arbitrary-scale strategy in the SR training process, enabling super-resolution of UHD images at any scale with a single trained model, and demonstrates the effectiveness and superiority of the proposed method.
- 1
- Automatic Prompt Generation and Grounding Object Detection for Zero-Shot Image Anomaly Detection1
A novel zero-shot training-free approach for automated industrial image anomaly detection using a multi-modal machine learning pipeline, consisting of three foundation models, that enables efficient, scalable, and objective quality control in industrial manufacturing settings.
- Aligning localization and classification for anchor-free object detection in aerial imagery1
This paper proposes a novel and efficient anchor-free object detector, namely localization-classification-aligned detector (LCA-Det), which explicitly pulls closer the predictions of localization and classification, through a single-branch subtask-aligned detection head and a subtask-aligned sample assignment metric.
- A Memory-Assisted Knowledge Transferring Framework with Curriculum Anticipation for Weakly Supervised Online Activity Detection1
This paper tackles the WS-OAD task from the knowledge-distillation (KD) perspective, which trains an online student detector to distill dual-level knowledge from a weakly supervised offline teacher model, and improves the vanilla KD framework from two aspects.
- SMART: stratified matching and recurrent transformer for optical flow estimation1
This paper presents a Transformer-based parallel refinement network to improve the accuracy of coarse predictions and allow fine predictions to be adjusted, based on the accurate coarse positional information.
- 1
- 1
- 1
- 1
- 1
- IA2U: A Transfer Plugin with Multi-Prior for In-Air Model to Underwater1
Experimental results show that IA2U combined with an in-air model can achieve superior performance in underwater image enhancement and object detection tasks, therefore considering the requirements of two different tasks, IE and OD.
- 1
- DWANet: Focus on Foreground Features for More Accurate Location1
Considering the importance of foreground features in the process of object detection, a foreground feature extraction module, based on deformable convolution, is proposed, and the attention mechanism is integrated to suppress the interference from the background.
- 1
- Face hallucination based on cluster consistent dictionary learning1
Experimental results demonstrate that the authors’ algorithm outperforms many state-of-the-art techniques for face hallucination under different datasets, based on cluster consistent dictionary learning with the assumption that human faces have similar facial structures.
- 1
- 1
- 1
- 1
- Can a machine have two systems for recognition, like human beings?1
Labels are divided into several hierarchies for efficient and accurate labeling, which are constructed using the Associative Memory Sharing method, proposed in this paper.
- Image classification without segmentation using a hybrid pyramid kernel1
A new hybrid pyramid kernel is proposed which incorporates local features extracted from both dense regular grids and interest points for image classification, without requiring segmentation, and has been proven to satisfy Mercer's condition and is efficient in measuring the similarity between image features.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- Semi-supervised Metric Learning for Image Classification1
This work proposes a semi-supervised extension to the Neighborhood Component Analysis (NCA) method, which is a supervised method especially tailored for KNN classifiers, and uses the learned distance metric to classify images using the KNN method.
- 1
- Image Restoration Based on PDEs and a Non-local Algorithm1
Experiments show that this method using the NL algorithm can improve the performance efficiently and significantly, and the proposed algorithm can take advantage of all the possible self-predictions provided by the image.
- 1
- 1
- Adaptive Alpha-Trimmed Average Operator Based on Gaussian Distribution Hypothesis Test for Image Representation1
A quadrangle collapse based mesh simplification algorithm which can rapidly produce high approximations of 3D models with minimal times of collapses than Garland's and Pan's algorithms is put forward.
- 1
- Face recognition based on eigen-illumination scheme and uncorrelated discriminant analysis1
A human face representation scheme that is insensitive to illumination variation is proposed in order to deal with varying illimunation problem for face recognition and an uncorrelated Linear Discriminant Analysis technique is proposed based on the eigen-illumination representation scheme.
- An effective dissolve detector using spatio-temporal slice1
This work proposes an efficient dissolve detector based on the spatio-temporal slices by using three simple second-order filters that can identify dissolves with arbitrary lengths accurately.
- 1
- 1
- 1
- 1
- An adaptive deformable template for mouth boundary modeling1
It will be shown that the role of priori knowledge of the domain is essential for perceptual organization in the algorithm and that model adaptation technique satisfactorily enhances the mouth boundary model in an automated fashion.
- –
- –
- Towards Robust Ergonomic Atssessment for Lifting Postures with Vision-Language Models–
A novel Vision–Language Model for lifting posture assessment (LPA-VLM) from a single image, which both perceives human body parts and leverages the reasoning ability of the Vision-language model (VLM) to analyze salient ergonomic determinants.
- MambaLIE: Scene Light Intensity-Boosted Low-Light Image Enhancement with State Space Model–
This work proposes MambaLIE, a Scene Light Intensity-Boosted Low-Light Image Enhancement method based on a State Space Model (SSM), which outperforms state-of-the-art CNN-based and Transformer-based LIE methods on four widely used synthetic benchmarks and five publicly available real-world benchmarks in terms of accuracy, speed, and model size.
- –
- –
- –
- –
- –
- –
- –
- Emotion Diffusion Classifier With Adaptive Margin Discrepancy Training for Facial Expression Recognition–
A conditional generative diffusion model is adopted and the Emotion Diffusion Classifier (EmoDC) is introduced for FER, which demonstrates enhanced adversarial robustness and AMDiT-enhanced EmoDC has better generalization and robustness than state-of-the-art discriminative classifiers.
- Degradation-Aware Prompt Learning With Cross-Modal Compensation for Adverse Weather Removal–
A Degradation-Aware Cross-Modal Prompt Compensation Network that leverages cross-modal degradation cues from a pre-trained vision-language model to condition restoration features within a unified backbone, achieving robust and perceptually consistent restoration across diverse weather conditions.
- –
- –
- –
- Multi-level distortion-aware deformable network for omnidirectional image super-resolution–
A novel Multi-level Distortion-aware Deformable Network (MDDN) for ODISR is proposed, designed to expand the sampling range and receptive field and generate dense and comprehensive features that effectively capture geometric distortions in ERP images.
- –
- –
- –
- –
- Toward effective local dimming-driven liquid crystal displays: a deep curve estimation–based adaptive compensation solution–
A deep lightweight network, namely, the curve estimation network (CENet), takes both the original input image and the dimmed BLUs as input, to estimate a set of high-order curves, and reformulate pixel compensation as an input-specific curve estimation task.
- –
- Dynamic spatial aggregation network for joint denoising and HDR imaging–
This paper proposes a dynamic spatial aggregation network, which can simultaneously handle noise and large motion of objects in LDR images in real-world scenarios and significantly outperforms state-of-the-art deep HDR models, resulting in high-quality HDR images in real-world situations.
- –
- An image segmentation algorithm for isolating ocean fronts of interest–
This paper proposes an image segmentation algorithm to segment ocean fronts that contributes to the advancement of oceanographic studies and supports improved understanding of the intricate dynamics within marine ecosystems.
- Label Text-aided Hierarchical Semantics Mining for Panoramic Activity Recognition–
A label text-aided hierarchical semantics mining (THSM) framework, which explores multi-level cross-modal associations by learning hierarchical semantic alignment between visual content and label texts and proves the superiority of the proposed THSM framework over state-of-the-art methods.
- –
- A robust and efficient automated product recognition system based on scenario design methodology and computer vision techniques–
The proposed approach speeds up the process by 200% to 300% for product matching, and the accuracy of product recognition in the front-view with rotations ranges from 90% to 95% using six different feature descriptors.
- Multi-level feature aggregation network for high dynamic range imaging–
The proposed multi-level feature aggregation network progressively aggregates non-overlapping frequency sub-bands at different pyramid levels, and generates the corresponding HDR image from coarser to finer scales, thereby producing HDR images with high visual quality.
- Visual-optical positioning and configuration of non-identified electronic pricing tags on retail shelves–
A new positioning methodology for electronic pricing tags (EPTs) is proposed, performed by a Visual-Optical Mobile Module (VOMM) on the shelf, which means that neither Radio Frequency Identification (RFID) nor a beacon is needed to tag the products.
- Deep residual convolutional neural network with curriculum learning for source camera identification–
This paper proposes to train an accurate deep residual convolutional neural network (ResNet), with the use of curriculum learning (CL) and preprocessed noise residues of camera images, so as to suppress contamination of camera fingerprints and extract highly discriminative features for camera identification.
- –
- –
- 3D model retrieval based on deep learning approach with weighted three-view deep features–
In this work, a new feature weighting algorithm is applied to the 3D-view-based features, which can emphasize the more important views of the3D models, so the retrieval performance can be improved.
- Elastic net with adaptive weight for image denoising–
An elastic net with adaptive weight for image denoising that can achieve nearly unbiased estimation and select highly correlated features simultaneously and outperforms other state-of-the-art image Denoising methods.
- Deep skip connection and multi-deconvolution network for single image super-resolution–
An efficient single image super-resolution (SR) method for multi-scale image texture recovery, based on Deep Skip Connection and Multi-Deconvolution Network is proposed, which achieves state-of-the- art performance.
- –
- Deep‐network based method for joint image deblocking and super‐resolution–
A joint deep-network-based image-restoration algorithm is proposed to establish a restoration framework for image deblocking and super-resolution and demonstrates that the proposed method outperforms other state-of-the-art methods, in terms of both subjective and objective performances.
- An Accelerated Procrustean Markov Process Model With Coherent Constraint for Non-Rigid Structure From Motion–
A coherent constraint is constructed to suppress the noise in the initialization step of thePMP algorithm, and an Accelerated Expectation Maximization (AEM) algorithm is devised to optimize the PMP estimation model.
- Point Set Registration With a Hybrid Structure Constraint–
This paper proposes an effective point set registration approach based on a hybrid structure constrain that is more robust due to the use of multiple effective constraints.
- Deeply learned pore-scale facial features with a large pore-to-pore correspondences dataset–
A constraint based on 3D facial model is employed and this dataset is then used to train a Convolutional Neural Network to generate the novel pore-scale facial features - Deeply Learned Pore- scale Facial Features (DLPFF).
- Object tracking with an effective composite similarity–
A robust object-tracking approach is proposed based on a composite similarity measure, which combines the likelihood value and the confidence value for each candidate and is a more robust approach to object tracking.
- –
- Accurate car plate detection via car face landmark localization–
This paper presents a novel, two-stage detection scheme for small, blurred car-plate detection in large surveillance images that firstly detects vehicles, and then locates the car plates in specific regions of detected vehicles based on the proposed car-face landmark localization algorithm.
- –
- –
- –
- –
- –
- An efficient approach for compound identification based on the frequency features of mass spectra–
An effective and efficient compound-identification approach is proposed based on the frequency features of mass spectra, where the computation time of the proposed method is only about 6%–11% of that of the existing methods, while the identification performance remains sufficiently competitive.
- –
- An Effective Approach for NRSFM of Small-Size Image Sequences–
A sub-sequence-based integrated algorithm is proposed to deal with the NRSFM problem with small sequence sizes that can achieve a higher estimation accuracy, and has better robustness.
- –
- –
- –
- –
- –
- A biased selection strategy for information recycling in Boosting cascade visual-object detectors–
This work proposes a biased selection strategy that promotes re-using existing information when selecting weak classifiers or features in each Boosting iteration, and solves the learning problem in a step-wise manner, similar to the gradient-Boosting scheme.
- A Mass Spectra-Based Compound-Identification Approach with a Reduced Reference Library–
A nonzero feature-retention strategy, and a correlation based-reference library reduction strategy, are designed in the proposed algorithm to reduce the computation burden.
- A Novel Ensemble Algorithm for Tumor Classification–
From the viewpoint of image processing, a spectral feature-based TLS (Tikhonov-regularized least-squares) ensemble algorithm is proposed for tumor classification using gene expression data, demonstrating the feasibility and effectiveness of the proposed method.
- –
- –
- –
- –
- –
- –
- Efficient rotation- and scale-invariant texture analysis–
A new set of kernels called Slit is proposed, which can replace Gabor wavelets in applications where high computational speed is desired and feature extraction using Slit requires only 11 to 17% of the numeric operations.
- –
- –
- Face recognition based on phase-feature–
A novel face recognition algorithm based on the histogram of the SGW phase is proposed that can achieve a better performance in terms of both recognition accuracy and tolerance to illumination variations.
- –
- –
- –
- Algorithm for visualization of high dynamic range images–
A novel algorithm based on spatial and statistical information was proposed for the display of HDR images to deal with the problems of overall impression preservation and visual details reproduction.
- –
- –
- –
- –
- –
- –
- An edge-preserved image denoising technique based on iterated function systems–
A new fractal-based image denoising method is proposed which can reserve the edges and remove the blocky artifacts in a denoised image using the decoupling property of the fractal code instead of the conventional fractal coding using the contrast scaling and offset parameters.
- –
- –
- An efficient method for facial expression recognition–
An efficient method for human facial expression recognition is presented that includes two major techniques: spatially maximum occurrence model (SMOM), which is used to describe the different facial expressions; and elastic shape-texture matching (ESTM), which was used to compute the similarity between two images.
- –
- –
- –
- Pulling, Pushing, and Grouping for Image Segmentation–
A novel computational visual grouping method, termed pulling, pushing and grouping, or PPG for short, based on the Hopfield neural model is presented for solving the optimisation process of visual grouping.
- Quality fine granular scalability video coding method for head shoulder sequence images–
A quality fine granular scalable coding scheme for head shoulder sequence images, which are commonly seen in video applications, is implemented and results show that the overall coding efficiency gain is higher than that of MPEG-4 FGS standard and the bit stream is finegranular scalable.
- –
- Finding Human Faces in a Face Database–
The different parts of a human face recognition system will be described, namely, locating human faces, extracting facial features, face recognition, and searching for faces from a database.
- –
- –
- Fast fractal-image compression using feature vector matching–
This paper presents a fast algorithm for fractal image compression that uses quadtree partitioning to partition an image into image blocks of different sizes and an efficient encoding approach for low complexity range blocks, which encodes the mean of a range block without searching the domain blocks.
- –
- –
- –
- –
- –
- –
- –
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 2026-10-15. 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.