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
- Jianping ZhangSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works178 from public data
- Video copy detection322
Local methods demonstrate their superior performance over the global ones, when detecting video copies subjected to various transformations, as well as for different lengths of video segments.
- 243
- Content Adaptive and Error Propagation Aware Deep Video Compression176
This work proposes a content adaptive and error propagation aware video compression system that outperforms the state-of-the-art learning based video codecs on benchmark datasets without increasing the model size or decreasing the decoding speed.
- 155
- Video sequence matching based on temporal ordinal measurement125
A novel video sequence matching method based on temporal ordinal measurements that gives a global and local description of temporal variation and a quantitative method to measure the robustness and discriminability attributes of the matching methods.
- 117
- Blurry Video Frame Interpolation111
A pyramid module to cyclically synthesize clear intermediate frames to reduce motion blur and up-convert frame rate simultaneously is developed and an inter-pyramid recurrent module is proposed to connect sequential models to exploit the temporal relationship.
- Graph-based topology reasoning for driving scenes78
TopoNet, an end-to-end topology reasoning network for analyzing driving scenes, is presented and evaluated on the challenging scene understanding benchmark, OpenLane-V2, where the approach outperforms all previous studies by a significant margin.
- Weld Defect Images Classification with VGG16-Based Neural Network72
This paper proposes a VGG16 based fully convolutional structure to classify the weld defect image, which achieves a high accuracy with a relative small dataset for deep learning method.
- 68
- 58
- 58
- 52
- GACV: Geodesic-Aided C–V method52
By practical experiments, it is verified that the proposed novel algorithm obtains better results than original methods, especially with respect to images within holes, complex background, weak edges, and noise.
- A Coding Framework and Benchmark Towards Low-Bitrate Video Understanding50
By introducing the attention mechanism and adaptive modeling scheme, the video semantic modeling ability of the approach is further enhanced, and a low-bitrate video understanding benchmark is built with three downstream tasks on eight datasets, demonstrating the notable superiority of the approach.
- 50
- 50
- 47
- Digital topological method for computing genus and the Betti numbers45
A linear time algorithm that determines topological invariants such as genus and the Betti numbers for digital spaces in 3D, based on cubical images with direct adjacency, could have applications in medical imaging as they can be used to identify patterns in3D image.
- Fuzzy subfiber and its application to seismic lithology classification42
The definition of fuzzy sub fiber is given and one of its most important properties: connectivity on fuzzy subfibers is discussed, which enables us to develop fast image segmentation algorithms for higher-dimensional range images.
- 41
- 39
- 39
- 36
- NTIRE 2021 Challenge on Video Super-Resolution36
This paper presents evaluation results from two competition tracks as well as the proposed solutions to the NTIRE 2021 Challenge on Video Super-Resolution, and develops conventional video SR methods focusing on the restoration quality.
- Edge guided salient object detection33
This paper derives two scales of nested superpixels from an edge guided segmentation and calculates a coarse saliency map on the finer scale by adopting color contrast, spatial prior and boundary prior, and proposes a novel background prior.
- The equivalence between two definitions of digital surfaces32
This paper presents a proof of the equivalence of two digital surface definitions based on simple surface points given by Morgenthaler and Rosenfeld and uses a parallel-move concept as given by direct adjacency.
- 30
- 26
- Local Label Point Correction for Edge Detection of Overlapping Cervical Cells24
The qualitative and quantitative experimental results show that the proposed local label point correction (LLPC) method can improve the quality of manual labels and the accuracy of overlapping cell edge detection.
- 23
- 22
- AIM 2019 Challenge on Video Temporal Super-Resolution: Methods and Results20
This paper reviews the first AIM challenge on video temporal super-resolution (frame interpolation) with a focus on the proposed solutions and results, and employs the REDS_VTSR dataset derived from diverse videos captured in a hand-held camera for training and evaluation purposes.
- PointSkelCNN: Deep Learning‐Based 3D Human Skeleton Extraction from Point Clouds18
This study converts 3D human skeleton extraction into offset vector regression and human body segmentation via deep learning‐based point cloud contraction and a disambiguation strategy is adopted to improve the robustness of joint points regression.
- 18
- 17
- 17
- 17
- HSPOG: An optimized target recognition method based on histogram of spatial pyramid oriented gradients16
This paper performs spatial pyramid segmentation on target images of any size, gets the pixel size of each image block dynamically, and further calculates and normalizes the gradient of the oriented feature of each block region in each image layer.
- 16
- 16
- Audio-Visual Saliency for Omnidirectional Videos15
This paper establishes the largest audio-visual saliency dataset for omnidirectional videos (AVS-ODV), which comprises the omniddirectional videos, audios, and corresponding captured eye-tracking data for three video sound modalities including mute, mono, and ambisonics, and compares the performance of several state-of-the-art saliency prediction models on the dataset.
- 15
- 15
- 15
- Gradually Varied Surfaces and Gradually Varied Functions15
An envelope theorem, a uniqueness theorem, and an extension theorem which concerns preserving the same norm, are obtained and it is shown that any digital manifold can normally immerse an arbitrary tree T.
- 14
- 14
- 13
- An Intra-BRNN and GB-RVQ Based END-TO-END Neural Audio Codec13
An end-to-end neural speech codec, namely CBRC (Convolutional and Bidirectional Recurrent neural Codec) is presented, designed to exploit the intra-frame correlations more efficiently and reduce the quantization noise.
- 13
- Alpha at SemEval-2021 Task 6: Transformer Based Propaganda Classification13
Two branches of techniques including fine-tuning the text pretrained transformer with extended visual features, and fine- Tuning the multimodal pretrained transformers are explored, which will help fuse the clues from both image and text.
- 13
- 13
- 13
- Blood quality evaluation via on-chip classification of cell morphology using a deep learning algorithm12
This method has the merits of cell identification in a wide channel, no need for single cell alignment as the image cytometry does and it is not only applicable to the quality evaluation of RBCs, but can also be used for general cell identifications with different morphologies.
- An acceleration-level visual servoing scheme for robot manipulator with IoT and sensors using recurrent neural network12
The proposed method extends the application scenarios of the traditional visual Servoing schemes by using acceleration commands, which also improves the safety of the visual servoing control of the redundant robot manipulator.
- 12
- A sparse robust model for large scale multi-class classification based on K-SVCR12
A robust least-squares version of K-SVCR (K-RLSSVCR) is proposed, which partially depress the impact of outliers on the new model via its nonconvex e-insensitive ramp loss and truncated least squares loss.
- 12
- 12
- 12
- Comprehensible volume LIC rendering based on 3D significance map12
Based on the 3D significance map, a 3D streamline illumination model is presented that can control transfer functions for comprehensible LIC volume rendering by highlighting significant regions and neglecting insignificant information.
- 11
- 11
- Adaptive Local Receptive Field Convolutional Neural Networks for Handwritten Chinese Character Recognition11
This work argues that the best way to select optimal local receptive field is to let CNNs learn how to choose it, and uses different sizes of local receptive fields to produce several sets of feature maps, then an element-wise max pooling layer is introduced.
- 11
- 11
- 10
- 9
- SwiftAvatar: Efficient Auto-Creation of Parameterized Stylized Character on Arbitrary Avatar Engines8
- 8
- Parallel-Friendly Patch Match Based on Jump Flooding8
A parallel-friendly algorithm for k-nearest neighbor based patch match based on jump flooding algorithm that achieves up to 100 times speedup over its CPU implementation, and 5 times faster than the GPU implementation of Barnes’s algorithm, a most recently benchmark algorithm.
- Multi-feature fusion method based on support vector machine and k-nearest neighbor classifier8
The traditional classification methods only use one single classifier, which may lead to one-sidedness, low accuracy, and that the samples nearby the Support Vector Machine(SVM) hyperplanes are more easily misclassified.
- Point spaces and raster spaces in digital geometry and topology8
The author's opinions on how to define digital curves, surfaces, and manifolds in point spaces or raster spaces are presented and some basic concepts in digital geometry and topology are unify.
- Classification of simple surface points and a global theorem for simple closed surfaces in three-dimensional digital spaces8
The corner point theorem is proved: Any simple closed surface has at least eight corner points, where a corner point of a closed surface is a point in the surface which has exactly three adjacent points in the closed surface.
- 7
- 7
- 7
- 7
- Multilevel manufacturing system of virtual enterprise based on manufacturing grid and strategies for member enterprise selection and task assignment7
An example of developing the artificial hip joint in thevirtual enterprise is used to demonstrate that efficiency of the virtual enterprise is improved by using the manufacturing grid and the proposed strategies for member enterprise selection and task assignment.
- 6
- 6
- 6
- 6
- 6
- 5
- 5
- 5
- 5
- Efficient Learning Based Sub-pixel Image Compression5
This framework follows the widely used variational auto-encoder architecture and is optimized based on the rate-distortion balance and is exploited to reduce the spatial resolution of image and improve the inference speed.
- Spatiotemporal salient object detection based on distance transform and energy optimization5
A novel spatiotemporal salient object detection method to produce high-quality saliency maps that can achieve good performance for complex scenes such as cluster backgrounds and non-rigid deformation is presented.
- 5
- lambda-Connectedness Determination for Image Segmentation5
This paper compares some segmentation technologies and attempts to find an automated way to better determine the parameters for image segmentation, especially the connectivity value of lambda in lambda-connected segmentation.
- 5
- 4
- Robust water meter reading recognition method for complex scenes4
In this method, the traditional water meter recognition method has been improved so that it can handle more variable real-world scenarios and a data augmentation method is proposed for half-word character recognition that is difficult to handle in the past, which greatly improve network performance.
- 4
- 4
- 4
- 4
- 4
- Pattern Recognition on Random Graphs4
A robust vertex classifier is proposed, which utilizes a representation of a test vertex as a sparse combination of the training vertices, and is demonstrated to be robust against data contamination, and has superior performance over classical spectralembedding classifiers in simulation and real data experiments.
- 4
- λ-Connectedness Determination for Image Segmentation4
This paper compares some segmentation technologies and attempts to find an automated way to better determine the parameters for image segmentation, especially the connectivity value of lambda in lambda-connected segmentation.
- An Attention Based Similarity Measure for Colour Images4
An attention based similarity measure in which only very weak assumptions are imposed on the nature of the features employed, which has the significant advantage that similarity matching is based on an unrestricted competition mechanism that is not dependent upon a priori assumptions regarding the data.
- Optimal Stealthy Attack With Side Information Under the Energy Constraint on Remote State Estimation3
This work considers the innovation-based attacks with side information under the energy constraint in cyber-physical systems where Kullback-Leibler (K-L) divergence is used as the stealthiness metric and finds the optimal attack policies for the TE and AE.
- 3
- 3
- 3
- 3
- 3
- Automatic movie restoration based on wave atom transform and nonparametric model3
An effective automatic movie restoration system based on wave atom transform and nonparametric model that achieves excellent performances even for seriously damaged movies but also has low computational complexity compared with the existing methods.
- New regularization scheme for blind color image deconvolution3
A reinforcement regularization framework that integrates a soft parametric learning term in addressing blind color image deconvolution and is able to achieve satisfactory restored color images under different blurring conditions is proposed.
- 3
- 3
- Image super-resolution using dilated neighborhood attention transformer2
A dilated neighborhood attention transformer for image SR (DiNAT-SR) is proposed to improve SwinIR for image SR and replaces SA with DiNA to capture more global data and allow the receptive field to grow exponentially.
- 2
- A deep encoder-decoder based primal-dual proximal network for image restoration2
A deep encoder-decoder primal-dual proximal network, named ED-PDPNet, which is shown to be a lightweight network with fewer learning parameters in comparison to the recent popular transformer-based method.
- Generative Adversarial Networks with Data Augmentation and Multiple Penalty Areas for Image Synthesis2
This research introduces a GAN architecture based on data augmentation, in order to make the model fulfill 1-Lipschitz constraints, it proposes to consider these additional data included into the penalty areas which can improve ability of discriminator and generator.
- A New Single Image Dehazing Method Based on Improved Double-area Filter and Guided Filter2
A novel transmission estimation method based on the improved doublearea filter and guided filter that not only provides better results for the depth-edge regions but also has lower complexity.
- 2
- 2
- Improved coded exposure for enhancing imaging quality and detection accuracy of moving targets2
An improved coded exposure is introduced in this paper to remove the image blur and obtain high quality image, so that the test accuracy of the surface defect and edge contour of motion objects can be enhanced.
- A Market Data Feeds Processing Accelerator Based on FPGA2
An accelerator based on FPGA for processing the market data feeds in OPRA format that focuses on encoding and decoding the data feeds concerning five of the most important categories, namely categories a, d, k, q and N.
- Object Tracking within the Framework of Concept Drift2
The proposed object tracking algorithm within the framework of concept drift improves the tracking robustness and accuracy which is illustrated by the two experiments on two real-world changing scenes.
- 2
- 2
- A Resource Model for Large-Scale Non-hierarchy Grid System2
This paper brings forward a resource model for future large-scale non-hierarchy grid system based on the assumption of power-law degree distribution and result of Lada A. Adamic, finding that the unstructured locating algorithms used in P2P system do not suit this resource model.
- Design and implementation of transport stream remultiplexer with cascade architecture2
The design and implementation of MPEG-2 transport stream (TS) remultiplexer with cascade architecture does not need the PCR correction procedure, and the synchronization of programs in the input TS can be preserved.
- Non-linear view interpolation2
A new non-linear view interpolation algorithm is presented that can exactly simulate the perspective viewing transformation during walkthrough and is much more accurate and efficient than the traditional one.
- Enhanced temporal encoding-decoding for survival analysis of multimodal clinical data in smart healthcare1
An enhanced temporal encoding-decoding (ETED) model that integrates an adaptive attention mechanism and a gated weighting mechanism to improve predictive performance is proposed, with comparative results demonstrating that the ETED variant incorporating facial features (ETEncoding-Decoding-Face) outperforms traditional models.
- 1
- 1
- 1
- Geodesic Learning With Uniform Interpolation on Data Manifold1
A method to learn a minimizing geodesic within a data manifold is proposed and is able to generate high-quality uniform interpolations with the shortest path between two given data samples.
- 3D Hand Pose Estimation for Guqin Performance1
3D hand pose surface estimation based on personalized hand features is proposed and applied to guqin performance and a new MMFPN structure is proposed, which can not only realize the three-dimensional surface estimation of basic finger-pointing, but also effectively solve the problem of self-occlusion.
- Method for Removing Motion Blur from Images of Harmful Biological Organisms in Power Places Based on Improved Cyclegan1
This work proposes a method for removing motion blur from the images of harmful biological organisms in power places by combining cycle consistency and perceptual loss to enhance the detail authenticity of image texture restoration and improve the detection accuracy.
- 1
- Survey of Pedestrian Action Recognition in Unmanned-Driving1
A detailed survey about the recent and state-of-the-art research methods in the filed of human action recognition and discusses their advantages and limitations, and analysis the main framework of motion recognition, and summarize the common datasets.
- 1
- 1
- A Novel Rate Control Algorithm for Low Complexity Frame Compression System1
This paper proposes a low complexity video compression implementation whose bitstreams are compatible with ProRes and can provide higher PSNR and less computing complexity than Kostya ProRes, meanwhile keeping comparable accuracy in rate control.
- 1
- Parallel-Friendly Frame Rate Up-Conversion Algorithm Based on Patch Match1
A parallel-friendly frame rate up-conversion (FRUC) algorithm based on the patch match that has a good performance in handling the fine motion structure and occlusion problem and can be implemented on GPU.
- 1
- Facial expression recognition based on image Euclidean distance-supervised neighborhood preserving embedding1
IED-SNPE not only the local structure of expression manifold preserves well but also explicitly considers the spatial relationships among pixels in the images, so it excels NPE in feature extraction and is highly competitive with those well-known feature extraction methods.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- Intelligent curve tracking algorithms and implementations1
A fuzzy neural network is proposed and implemented for tracking curves in sequential images that includes the fuzzification of images, fuzzy segmentation, sub-curve search, and genetic algorithm linking.
- Regular polygons and their application to digital curves1
This note discusses various kinds of 2D unit cells, or surface-unit cells, made by regular polygons (simplexes) in the plane R X R, and gives a type of topological proofs to that a closed digital curve has at least 6 points in a 3-regular-polygon decomposition plane.
- Noise and shuffle feature perturbation driven robust curve segmentation–
A lightweight feature perturbation mechanism that enables flexible integration with existing segmentation models and incurs no additional computational complexity overhead is proposed, and has made certain improvements over the existing state-of-the-art methods on multiple standard datasets.
- T-HITL Effectively Addresses Problematic Associations in Image Generation and Maintains Overall Visual Quality–
A conceptual and technical taxonomy for addressing problematic associations in machine learning models and generative AI and a method, T-HITL, that addresses these associations and simultaneously maintains visual quality of image model generations is provided.
- –
- –
- –
- –
- –
- Automatic Measurement of Blood Vessel Angles in Immunohistochemical Images of Liver Cancer–
This system is the first one that conducts automated measurement of blood vessel angle of immunohistochemistry and results in better accuracy, according to the experts’ results.
- –
- –
- –
- –
- –
- Adaptive block size selection for inter-layer backward view synthesis prediction–
A block based BVSP for inter-layer prediction with only high-level syntax changes is implemented and an adaptive depth block size selection method is proposed and the experimental results show that an average gain of 3.5% bitrate reduction was achieved and this performance gain is relatively maintained while the number of motion compensation operations was reduced to a designated level.
- Education forum–
This paper gives a brief introduction to digital geometry and its relationships to algorithm analysis, image processing, and computer graphics and suggests that digital geometry can be taught as a chapter in computational geometry or geometric methods for computing sciences courses.
- Monocular Vision Based Relative Depth Estimation for Hand Gesture Recognition–
This work focuses on real-time relative hand depth estimation with cluttered backgrounds and variable illumination for a target application to hand pushing and pulling detection and adopts a detection-and-registration strategy to predict frame-toframe scaling factor and make decision by the cumulated relative scale changes.
- –
- –
- –
- A Novel Color C-V Method and Its Application–
A novel color C-V method which takes into account of color information and global property is presented and the experimental results show robust performance especially in case of weak edges and noisy inputs.
- An improved QTCQ wavelet image coding method using DCT and coefficient reorganization–
An improved quadtree classification and TCQ (QTCQ) wavelet image compression method that applies small block DCT to coefficients in the high frequency subbands and reorders them before quantizing and coding.
- –
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.