Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks117 from public data
- 1,015
- 271
- HVI: A New Color Space for Low-light Image Enhancement244
A new color space for LLIE is proposed, namely Horizontal/Vertical-Intensity (HVI), defined by polarized HS maps and learnable intensity, defined by polarized HS maps and learnable intensity.
- NTIRE 2024 Challenge on Low Light Image Enhancement: Methods and Results85
This paper meticulously evaluates the state-of-the-art advancements in enhancing low-light images, reflecting the significant progress and creativity in this field.
- 84
- ACDMSR: Accelerated Conditional Diffusion Models for Single Image Super-Resolution73
This research revisits diffusion models in image super-resolution and proposes a straightforward yet significant diffusion model-based super-resolution method called ACDMSR (accelerated conditional diffusion model for image super-resolution), which surpasses previous attempts in qualitative and quantitative results.
- High dynamic range imaging by sparse representation72
A ghost-free HDR image synthesis algorithm that utilizes a sparse representation framework based on the dependency among adjacent low dynamic range images and the sparsity of the moving object that leads to the ghost artifacts is proposed.
- GoMVS: Geometrically Consistent Cost Aggregation for Multi-View Stereo53
GoMVS is proposed to aggregate geometrically consis-tent costs, yielding better utilization of adjacent geometries by leveraging the local geomet-ric smoothness in conjunction with surface normals by the geometric consistent propagation (GCP) module.
- 53
- Learning to Fuse Monocular and Multi-view Cues for Multi-frame Depth Estimation in Dynamic Scenes53
A novel method to learn to fuse the multi-view and monocular cues encoded as volumes without needing the heuristically crafted masks is proposed, which includes the cross-cue attention (CCA) to encode the spatially non-local relative intra-relations from each source to enhance the representation of the other.
- Learning depth via leveraging semantics: Self-supervised monocular depth estimation with both implicit and explicit semantic guidance50
A Semantic-aware Spatial Feature Alignment (SSFA) scheme to effectively align implicit semantic features with depth features for scene-aware depth estimation and a semantic-guided ranking loss to explicitly constrain the estimated depth maps to be consistent with real scene contextual properties are proposed.
- A Unified HDR Imaging Method with Pixel and Patch Level50
A hybrid HDR deghosting network, called HyHDRNet, is proposed to learn the complicated relationship between reference and non-reference images and outperforms state-of-the-art methods both quantitatively and qualitatively, achieving appealing HDR visualization with unified textures and colors.
- Attention-Based Network For Low-Light Image Enhancement48
A novel attention-based neural network to generate high-quality enhanced low-light images from the raw sensor data and proposes a new pooling layer, called inverted shuffle layer, which adaptively selects useful information from previous features.
- 46
- 45
- All-in-one Multi-degradation Image Restoration Network via Hierarchical Degradation Representation42
A novel All-in-one Multi-degradation Image Restoration Network (AMIRNet) that can effectively capture and utilize accurate degradation representation for image restoration and design a feature transform block (FTB) that aligns domains and refines features with the guidance of the degradation representation.
- High dynamic range imaging via gradient-aware context aggregation network38
The proposed HDR imaging approach that aggregates the information from multiple LDR images with guidance from image gradient domain generates artifact-free images by integrating the image gradient information and the image context information in the pixel domain.
- Going the Extra Mile in Face Image Quality Assessment: A Novel Database and Model34
The largest annotated IQA database developed to date is introduced, which contains 20,000 human faces – an order of magnitude larger than all existing rated datasets of faces – of diverse individuals in highly varied circumstances and a novel deep learning model is proposed to accurately predict face image quality.
- 32
- 32
- Sparse2DGS: Geometry-Prioritized Gaussian Splatting for Surface Reconstruction from Sparse Views30
This work proposes Sparse2DGS, an MVS-initialized Gaussian Splatting pipeline for complete and accurate reconstruction, which outperforms existing methods by notable margins while being 2× faster than the NeRF-based fine-tuning approach.
- Ghost Removal via Channel Attention in Exposure Fusion29
A novel Multi-scale Channel Attention guided Network (MCANet) is proposed to address the ghosting problem by using multi-scale blocks consisting of dilated convolution layers to extract informative features.
- Dim small target detection based on convolutinal neural network in star image27
A fully convolutional neural network is proposed to achieve pixel-wise classification, which can complete target-background separation in a single stage rapidly and outperforms the state-of-the-art in terms of higher detection rate and less false alarm caused by noises.
- 26
- Boosting HDR Image Reconstruction via Semantic Knowledge Transfer25
This framework introduces the Semantic Priors Guided Reconstruction Model (SPGRM), which leverages SDR image semantic knowledge to address ill-posed problems in the initial HDR reconstruction results, and leverages a self-distillation mechanism that constrains the color and content information with semantic knowledge, aligning the external outputs between the baseline and SPGRM.
- 25
- Enhancing image visuality by multi-exposure fusion24
A novel multi-exposure fusion based visuality enhancement method based on the Laplacian pyramid based linear fusion framework, which is able to simultaneously highlight pixels with good exposedness, contrast, saturation as well as Gamut, and well highlights the fine details.
- SMAE: Few-shot Learning for HDR Deghosting with Saturation-Aware Masked Autoencoders23
This work proposes a novel semi-supervised approach to realize few-shot HDR imaging via two stages of training, called SSHDR, and designs a Saturated Mask AutoEncoder (SMAE) to learn a robust feature representation and reconstruct a non-saturated HDR image.
- 22
- 22
- Video super-resolution via dense non-local spatial-temporal convolutional network20
A novel end-to-end deep neural network that can effectively model the sudden and smooth varying motions of videos and overcome the limitations of explicit motion estimation is presented.
- Non-uniform motion deblurring with blurry component divided guidance19
A deep two-branch network is presented to deal with blurry images via a component divided module, which divides an image into two components based on the representation of blurry degree, which performs favorably against the state-of-the-art approaches.
- Exploring and Evaluating Image Restoration Potential in Dynamic Scenes15
This paper first establishes a dynamic scene imaging dataset containing composite distortions and applied image restoration processes to validate the rationality of the existence to IRP, and proposes a novel deep model to accurately predict IRP values.
- Enhancing Self-supervised Monocular Depth Estimation via Incorporating Robust Constraints15
A gradient-based robust photometric loss is proposed which restrains the false supervisory signals caused by brightness changes and constrain the motion estimation network to generate across-frame consistent motions via proposing a triplet-based cycle consistency constraint.
- ARSAC: Efficient model estimation via adaptively ranked sample consensus15
A novel Adaptively Ranked Sample Consensus (ARSAC) algorithm is presented to boost the speed and robustness of RANSAC, and adopts non-uniform sampling based on the ranked measurements to speed up the sampling process.
- 14
- Modeling optical imaging pipeline and learning contrastive-based representation for hybrid-corrupted image restoration13
This work introduces Opticformer, a Transformer-based model adept at adaptively focusing on critical regions and addressing spatially varying corruptions, and incorporates a novel contrastive-based representation learning method, enabling Opticformer to generate multi-scale refined representations without prior knowledge.
- A multi-scale feature cross-dimensional interaction network for stereo image super-resolution13
A multi-scale feature cross-dimensional interaction network (MFCINet) for SSR is proposed, which better leverages intra-view and inter-view information to reconstruct clear texture details, achieving competitive results and outperforming state-of-the-art methods.
- 13
- 12
- GRAN: ghost residual attention network for single image super resolution11
This paper introduces Ghost Residual Attention Block (GRAB) groups to overcome the drawbacks of the standard convolutional operation, i.e., redundancy of the intermediate feature.
- 11
- C-Drag: Chain-of-Thought Driven Motion Controller for Video Generation10
A Chain-of-Thought-based motion controller for controllable video generation, named C-Drag, which first performs object perception and then reasons the dynamic interactions between different objects according to the given motion control of the objects.
- Multiple Object Tracking Based on Occlusion-Aware Embedding Consistency Learning10
A novel multiple object tracking method based on visual embedding consistency that is capable of addressing track interruptions caused by occlusion in online tracking scenarios and achieves promising performance levels in both unoccluded and occluded tracking scenarios.
- 10
- 10
- Dynamic Negative Sampling Autoencoder for Hyperspectral Anomaly Detection10
A dynamic negative sampling autoencoder is proposed for the hyperspectral anomaly detection (DNA-HAD), which is not only robust to reconstruct the background but also sensitive to detect the anomalies.
- SlimSeg: Slimmable Semantic Segmentation with Boundary Supervision10
This paper proposes a simple but effective slimmable semantic segmentation (SlimSeg) method, which can be executed at different capacities during inference depending on the desired accuracy-efficiency tradeoff, and employs parametrized channel slimming by stepwise downward knowledge distillation during training.
- Blind image deblurring by promoting group sparsity10
The proposed group sparsity priors are based on the fact that the nonzero elements of natural image gradients and blur kernels tend to cluster in structured group pattern and are introduced, which preserve the salient structures and smooth the minor components in imageGradients and restrict the blur kernel in a domain of dynamic group sparse vector.
- 9
- Multi-Granularity Language-Guided Training for Multi-Object Tracking8
It is argued that multi-modal language-driven features provide complementary information to classical visual features, thereby aiding in improving the robustness to such environmental interference and exhibiting strong cross-domain generalizability.
- 8
- Enhancing Real-Time Aerial Image Object Detection with High-Frequency Feature Learning and Context-Aware Fusion8
Extensive experiments and ablation studies on three public aerial image datasets validate that the proposed VMC-DETR framework effectively balances accuracy and computational efficiency, consistently outperforming state-of-the-art methods.
- Motion deblurring using the similarity of the multiscales8
Experiments indicate that the proposed blind image deblurring method could obtain better performance than the state-of-the-art methods, and could obtain satisfyingdeblurring result with fewer artifacts.
- 7
- 7
- 6
- Robust Motion Blur Kernel Estimation by Kernel Continuity Prior6
Inspired by the fact that the blur kernel is highly related to the continuous camera motion trajectory during the image capturing, it is proposed to encourage the continuity of the kernel through a kernel prior that outperforms state-of-the-art methods on both the normal and challenging cases.
- 5
- GSDD: Generative Space Dataset Distillation for Image Super-resolution5
Experimental results demonstrate that when trained with distilled data, GSDD can achieve comparable performance to the state-of-the-art (SOTA) SISR algorithms, while a nearly ×8 increase in training efficiency and a saving of almost 93.2% data storage space can be realized.
- UIR-LoRA: Achieving Universal Image Restoration through Multiple Low-Rank Adaptation5
This work proposes a universal image restoration framework based on multiple low-rank adapters (LoRA) from multi-domain transfer learning that leverages the pre-trained generative model as the shared component for multi-degradation restoration and transfers it to specific degradation image restoration tasks using low-rank adaptation.
- Take a Prior from Other Tasks for Severe Blur Removal5
This work proposes a cross-level feature learning strategy based on knowledge distillation to learn the priors, which include global contexts and sharp local structures for recovering potential details in general scene deblurring cases.
- 5
- Target recognition initiated self-dissociation based DNA nanomachine for sensitive and accurate MicroRNA (miRNA) detection5
A novel simple and sensitive miRNA detection approach by exploiting an elegantly designed target recognition initiated self-dissociation based DNA nanomachine that exhibits a high sensitivity with the limit of detection as low as 12 fM.
- 5
- Blur kernel estimation of noisy-blurred image via dynamic structure prior5
Comprehensive experimental results show that the proposed dynamic structure prior (DSP) outperforms previous methods in both commonly used datasets with various noise levels and real world images, from monochrome image to color image.
- Complementary coded aperture set for compressive high-resolution imaging5
A criterion for evaluating compressive coded aperture set with respect to the spectral complementarity maximization and the precision of restoration is derived and a genetic algorithm is optimized to get a set of coded aperture which can preserve more high frequency part together when compressivecoded sampling.
- 5
- A Dim Small Target Detection Method Based on Spatial-Frequency Domain Features Space5
A dim small target detection method based on the high dimensional spatial-frequency domain features extracted by filter bank, and training the support vector machine (SVM) classifier that exceeds the state-of-the-art on the ability to detect low SNR targets.
- 5
- 4
- Enhancing Border Learning for Better Image Denoising3
This work extends the patch-wise autoencoder to learn image mappings through patch extraction and patch-averaging operations, demonstrating that the patch-wise autoencoder is equivalent to a specific convolutional neural network (CNN) architecture, resulting in a novel residual block, referred to as the Border-Enhanced Residual Block (BERBlock).
- 3
- 3
- A Novel Method for Smearing Intensity Estimation and Elimination3
Experimental results of real images show that this method can effectively remove the smear signal while better preserve characteristics of weak target.
- 3
- A new method of moving object detection and shadow removing3
An adaptive method of objects and shadows detection in video streams that is adaptively updated in Hue Saturation Intensity color space to detect motion regions and remove cast shadows by the generic properties of luminance, chrominance and gradient density.
- 2
- Boosting Multi-view Stereo with Late Cost Aggregation2
This work presents a late aggregation approach that allows for aggregating pairwise costs throughout the network feed-forward process, achieving accurate estimations with only minor changes of the plain CasMVSNet.
- 2
- 2
- 2
- Super-Resolution of Video Using Deformable Patches2
This work proposes the concept that the high frequency video details could be estimated from the external examples effectively and shows that the method outperforms the state-of-art methods.
- 2
- An Image Retrieval Method Based on Region2
A new algorithm based on region is proposed that implements similarity measurement of two images by taking into consideration the similarity of every region synthetically and can work effectively and efficiently.
- DifAugGAN: A Practical Diffusion-style Data Augmentation for GAN-based Single Image Super-resolution1
This work proposes a simple but non-travel diffusion-style data augmentation scheme for current GAN-based SR methods, known as DifAugGAN, which involves adapting the diffusion process in generative diffusion models for improving the calibration of the discriminator during training motivated by the successes ofData augmentation schemes in the field to achieve good calibration.
- Blind Image Quality Assessment via Deep Recursive Convolutional Network with Skip Connection1
A multi-scale recursive deep neural network to accurately predict image quality and using skip connection to combine information among layers, and it further enriches the coarse and fine features for quality assessment.
- ARSAC: Robust Model Estimation via Adaptively Ranked Sample Consensus1
An adaptive scheme which updates the ranking of the measurements on each trial, to incorporate high quality measurement into sample at high priority and design a geometric constraint during sampling process, which could alleviate degenerate cases caused by non-uniform sampling in epipolar geometry.
- Blind Multi-frame Super Resolution with Non-identical Blur1
An novel blind multi-frame super resolution method with non-identical blur that can obtain superior results, and outperforms the state of the art in the experiments through subjective and objective evaluation is proposed.
- 1
- 1
- Computational imaging technology for high-sensitivity space-image acquisition1
A novel computational imaging method for high-sensitivity image acquisition that introduces CI into image formation and conducts minor modifications for cameras to encode more desirable information during capture, which is practical for hardware implementation.
- 1
- 1
- 1
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Multi-image Deblurring Using Complement–
A blurry images restoration method based on the complement is proposed for the selection of images in multi-image restoration, and an existing deblurring method is used to estimate the latent image.
- Statistical learning modeling method for space debris photometric measurement–
A statistical learning modeling method for space debris photometric measurement is proposed based on the global consistency of the star image, and the statistical information of star images is used to eliminate the measurement noises.
- –
- Dim Target Tracking Base on GM-PHD Filter–
Extensive experiments on real images sequences show that the proposed approach could effectively meet the requirements of the real-time detection with a low false alarm rate and a high detection probability.
- Small space target detection based on feature stability algorithm–
Qualitative and quantitative results prove that the proposed algorithm has strong anti-noise performance and fit for kinds of working pattern of detection system for target detection conveniently.
- Smear radiometric correction algorithm in star images based on kernel density estimation–
Experimental results indicate that the algorithm in this paper can remove smear effect in star image efficiently while retaining origin information and an algorithm to do radiometric correction in smear images based on modeling and estimating the probability density function of background noise in star images.
- Research of real time dim target detection algorithm under occlusion in space background–
Simulation results show that the proposed approach could effectively meet the requirements of the real-time detection with a low false alarm rate and a high detection probability, and it can also create a accuracy centroid when occlusion happens.
- Small and dim point target detection in special feature space–
The experimental result shown that the algorithm can fully take into account the characteristics of the fixed star, dim point moving target and noise, and can be effective to detect the moving dim and small in moving background with low SNR.
- Small and dim target detection based on selective visual attention mechanism and curve detection method–
The experiments proved that the suggested algorithm reduced significantly the processed information, demonstrated a strong anti- noise performance, and was suitable to detect the small and dim target in the low signal-to-noise environment.
- Edge Detection Method Based on Linear Singularity of Image–
Directional Beamlet Transform (DBT) is better than traditional Beamlet transform about directional information of image and has the ability of representing linear-singularity of image.
- Trajectory detection by nonuniform quantitative Hough Transform in segmented blocks–
The experiment results show that the nonuniform quantitative Houghtransform algorithm proposed inherits the merits of traditional Hough transform, and overcomes the disadvantages mentioned above, and detects dim and small targets efficiently in the image sequence of star field.
- A method of extracting airfield runway in remote sensing image–
The experimental results of remote sensing images show that the proposedMultiscale Hough transform combining Houghtransform and the partition of image space,olves the problem of location of line segments and improves the ability to detect short line segments.
- –
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.