A5003969300- 3D Shape Modeling and Analysis
- Computer Graphics and Visualization Techniques
- Computational Geometry and Mesh Generation
- Advanced Numerical Analysis Techniques
- Advanced Vision and Imaging
- 3D Surveying and Cultural Heritage
- Robotics and Sensor-Based Localization
- Image Processing and 3D Reconstruction
- Generative Adversarial Networks and Image Synthesis
- Advanced Image and Video Retrieval Techniques
- Advanced Image Processing Techniques
- Manufacturing Process and Optimization
- Face recognition and analysis
- Digital Media Forensic Detection
- Image and Object Detection Techniques
- Remote Sensing and LiDAR Applications
- Image Enhancement Techniques
- Image Retrieval and Classification Techniques
- Handwritten Text Recognition Techniques
- Data Visualization and Analytics
- Video Analysis and Summarization
- Human Motion and Animation
- Color Science and Applications
- Image and Signal Denoising Methods
- Optical measurement and interference techniques
Chinese Academy of Sciences
2016-2025
Beijing Academy of Artificial Intelligence
2020-2025
Shandong Institute of Automation
2012-2025
Institute of Automation
2016-2025
University of Chinese Academy of Sciences
2018-2025
Chengdu University of Technology
2024
Southwest University of Science and Technology
2024
Tsinghua University
2004-2023
Shuguang Hospital
2023
Shanghai University of Traditional Chinese Medicine
2023
Centroidal Voronoi tessellation (CVT) is a particular type of that has many applications in computational sciences and engineering, including computer graphics. The prevailing method for computing CVT Lloyd's method, which linear convergence inefficient practice. We develop new efficient methods computation demonstrate the fast these methods. Specifically, we show energy function 2nd order smoothness convex domains with smooth density, as well most situations encountered optimization. Due to...
Image inpainting has made remarkable progress with recent advances in deep learning. Popular networks mainly follow an encoder-decoder architecture (sometimes skip connections) and possess sufficiently large receptive field, i.e., larger than the image resolution. The field refers to set of input pixels that are path-connected a neuron. For task, however, size surrounding areas needed repair different kinds missing regions different, very is not always optimal, especially for local...
Abstract We propose a new isotropic remeshing method, based on Centroidal Voronoi Tessellation (CVT) . Constructing CVT requires to repeatedly compute Restricted Diagram (RVD) , defined as the intersection between 3D diagram and an input mesh surface. Existing methods use some approximations of RVD. In this paper, we introduce efficient algorithm that computes RVD exactly robustly. As consequence, achieve better quality than approximation‐based approaches, without sacrificing efficiency. Our...
Large-scale acquisition of exterior urban environments is by now a well-established technology, supporting many applications in search, navigation, and commerce. The same is, however, not the case for indoor environments, where access often restricted spaces are cluttered. Further, such typically contain high density repeated objects (e.g., tables, chairs, monitors, etc.) regular or non-regular arrangements with significant pose variations articulations. In this paper, we exploit special...
Distinguishing between natural images (NIs) and computer-generated (CG) by naked human eyes is difficult. In this paper, we propose an effective method based on a convolutional neural network (CNN) for fundamental image forensic problem. Having observed the rather limited performance of training existing CCNs from scratch or fine-tuning pre-trained network, design implement new appropriate with two cascaded layers at bottom CNN. Our can be easily adjusted to accommodate different sizes input...
Good building layouts are required to conform regulatory guidelines, while meeting certain quality measures. While different methods can sample the space of such good layouts, there exists little support for a user understand and systematically explore samples. Starting from discrete set we analytically characterize local shape around each initial layout, compactly encode these spaces, link them transitions across spaces. We represent in form portal graph. The then use graph, along with...
One-shot video-driven talking face generation aims at producing a synthetic video by transferring the facial motion from to an arbitrary portrait image. Head pose and expression are always entangled in transferred simultaneously. However, entanglement sets up barrier for these methods be used editing directly, where it may require modify only while maintaining unchanged. One challenge of decoupling is lack paired data, such as same but different expressions. Only few attempt tackle this with...
How things work visualizations use a variety of visual techniques to depict the operation complex mechanical assemblies. We present an automated approach for generating such visualizations. Starting with 3D CAD model assembly, we first infer motions individual parts and interactions between based on their geometry few user specified constraints. then this information generate that incorporate motion arrows, frame sequences animation convey causal chain parts. results wide
In this paper, we address the following research problem: How can generate a meaningful split grammar that explains given facade layout? To evaluate if is meaningful, propose cost function based on description length and minimize using an approximate dynamic programming framework. Our evaluation indicates our framework extracts grammars are competitive with those of expert users, while some users all competing automatic solutions less successful.
We present a novel and efficient approach to estimate 6D object poses of known objects in complex scenes represented by point clouds. Our is based on the well-known pair feature (PPF) matching, which utilizes self-similar pairs compute potential matches thereby cast votes for pose voting scheme. The main contribution this paper an improved PPF-based recognition framework, especially new center strategy relative geometric relationship between features. Using relationship, we first generate...
Reverse engineering CAD models from raw geometry is a classic but strenuous research problem. Previous learning-based methods rely heavily on labels due to the supervised design patterns or reconstruct shapes that are not easily editable. In this work, we introduce SECADNet, an end-to-end neural network aimed at reconstructing compact and easy-to-edit in self-supervised manner. Drawing inspiration modeling language most commonly used modern software, propose learn 2D sketches 3D extrusion...
Gliomas are the most aggressive type of malignant brain tumors.Recent studies have demonstrated that existence glioma stem cells (GSCs) is critical for recurrence, metastasis, and chemo-or radio-therapy resistance.Temozolomide (TMZ) has been used as an initial therapy gliomas.However, overall survival time still limiting due to lack effective targets treatment options.Therefore, identifying novel biomarkers gliomas, especially GSCs, important improve clinical outcome in future.In this study,...
In this article, we study the generation of maximal Poisson-disk sets with varying radii. First, present a geometric analysis gaps in such disk sets. This is basis for and adaptive sampling Euclidean space on manifolds. Second, propose efficient algorithms data structures to detect update when disks are inserted, deleted, moved, or their radii changed. We build concepts regular triangulations power diagram. Third, show how our contributes state-of-the-art surface remeshing.
Surface remeshing is a key component in many geometry processing applications. The typical goal consists finding mesh that (1) geometrically faithful to the original geometry, (2) as coarse possible obtain low-complexity representation and (3) free of bad elements would hamper desired application (e.g., minimum interior angle above an application-dependent threshold). Our algorithm designed address all three optimization goals simultaneously by targeting prescribed bounds on approximation...
In this paper, we describe a novel procedural modeling technique for generating realistic plant models from multi-view photographs. The realism is enhanced via visual and spatial information acquired images. contrast to previous approaches that heavily rely on user interaction segment plants or recover branches in images, our method automatically estimates an accurate depth map of each image extracts 3D dense point cloud by exploiting efficient stereophotogrammetry approach. Taking as soft...
Specular reflections pose great challenges on various multimedia and computer vision tasks, <i>e.g.</i>, image segmentation, detection matching. In this paper, we build a large-scale Paired Specular-Diffuse (PSD) dataset, where the images are carefully captured by using real-world objects ground-truth specular-free diffuse provided. To best of our knowledge, is first benchmark dataset for specular highlight removal task, which useful evaluating encouraging new deep learning-based approaches....
Abstract Specular highlight detection and removal is a fundamental problem in computer vision image processing. In this paper, we present an efficient end-to-end deep learning model for automatically detecting removing specular highlights single image. particular, encoder—decoder network utilized to detect highlights, then novel Unet-Transformer performs removal; append transformer modules instead of feature maps the Unet architecture. We also introduce module as mask guide task. Thus, these...