A5017302710- Computer Graphics and Visualization Techniques
- 3D Shape Modeling and Analysis
- Advanced Vision and Imaging
- Advanced Numerical Analysis Techniques
- Environmental and Agricultural Sciences
- Remote Sensing and Land Use
- Computational Geometry and Mesh Generation
- Environmental Changes in China
- Human Pose and Action Recognition
- 3D Surveying and Cultural Heritage
- Medical Image Segmentation Techniques
- Image Retrieval and Classification Techniques
- Generative Adversarial Networks and Image Synthesis
- Face recognition and analysis
- Viral Infections and Immunology Research
- Cardiovascular Effects of Exercise
- Data Management and Algorithms
- Digital Image Processing Techniques
- Advanced Image and Video Retrieval Techniques
- Cardiomyopathy and Myosin Studies
- Advanced Topics in Algebra
- Data Stream Mining Techniques
- Climate variability and models
- Advanced Clustering Algorithms Research
- Advanced Image Fusion Techniques
Hangzhou Dianzi University
2010-2023
Ministry of Education of the People's Republic of China
2019-2020
Chinese Academy of Meteorological Sciences
2011
Zhejiang University
2006
Abstract This paper presents a novel modeling system, called B‐Mesh, for generating base meshes of 3D articulated shapes. The user only needs to draw one‐dimensional skeleton and specify key balls at the skeletal nodes. system then automatically generates quad dominant initial mesh. Further subdivision evolution are performed refine mesh generate which has good edge flow along directions. can also modify manipulate shape by editing easily compose new shapes cutting pasting existing models in...
We present a novel solution to enable portrait relief modeling from single image. The main challenges are geometry reconstruction, facial details recovery and depth structure preservation. Previous image-based methods developed for bas-relief in 2.5D form, but not adequate 3D-like high with undercut features. In this paper, we propose template-based framework generate reliefs of various forms. Our method benefits Shape-from-Shading (SFS). Specifically, use bi-Laplacian mesh deformation guide...
This paper presents a novel approach for surface smoothing with feature preservation on arbitrary meshes. Laplacian operator is performed in global way over the mesh. The formulated as quadratic optimization problem, which easily solved sparse linear system. cost function to be optimized penalizes deviations from while maintaining overall shape of original features mesh can preserved by adding constraints and barycenter Our simple, non-iterative, fast, does not cause shrinkage distortion....
Background: Ventricular arrhythmias are associated with sudden cardiac death (SCD) in hypertrophic cardiomyopathy (HCM). Previous studies have found the late gadolinium enhancement (LGE) on magnetic resonance (CMR) was independently ventricular arrhythmia (VA) HCM. The risk stratification of VA remains complex and LGE is present majority HCM patients. This study conducted to determine whether scar heterogeneity from LGE-derived entropy VAs Materials Methods: Sixty-eight patients scarring...
3D content creation is referred to as one of the most fundamental tasks computer graphics. And many modeling algorithms from 2D images or curves have been developed over past several decades. Designers are allowed align some conceptual sketch suggestive curves, front, side, and top views, then use them references in constructing a model automatically manually. However, best our knowledge, no studies investigated on human body reconstruction similar manner. In this paper, we propose deep...
<p indent=0mm>An important part of autonomous driving is the perception environment car, which has created a strong demand for high precision semantic segmentation algorithms that can be run in real time on low-power mobile devices. However, when analyzing factors affect accuracy and speed network, it found structure previous algorithm, spatial information context features are difficult to take into account at same time, using two networks obtain separately will increase amount calculation...
3D CAD model retrieval plays a fundamental role in producing impressive and innovative designs easier faster. However #after decades of efforts, effective approach for furniture models is still rare. A novel models, based on hierarchical layout deduction, proposed according to the common structure characteristics furniture. In order discriminate effectively both topological, geometrical semantic layers, special components are defined multiresolution descriptor, attributed hierarchy graph,...
This paper aims at extending the method of Zhang et al. [1] to produce not only portrait bas-reliefs from single photographs, but also multi-style reliefs with adjustable depth range and reasonable ordering. To this end, we first propose an optimization-based synthesize high-depth 3D portraits. Then, train a normal-to-depth network learn mapping normal maps relief depths. After that, use trained construct samples given [1]. As each map has pixel-wise photograph, are able make correspondences...
This paper aims at extending the method of Zhang et al. (2023) to produce not only portrait bas-reliefs from single photographs, but also high-depth reliefs with reasonable depth ordering. We cast this task as a problem style-aware photo-to-depth translation, where input is photograph conditioned by style vector and output relief desired style. To construct ground-truth data for network training, we first propose an optimization-based synthesize 3D portraits. Then, train normal-to-depth...
In this paper, we present an end-to-end neural solution to model portrait bas-relief from a single photograph, which is cast as problem of image-to-depth translation. The main challenge the lack data for network training. To solve problem, propose semi-automatic pipeline synthesize samples. idea first construct normal maps photos, and then generate samples by reconstructing pixel-wise depths. total, our synthetic dataset contains 23 k photo/bas-relief pairs. Since process synthesis requires...