A5101618860- 3D Shape Modeling and Analysis
- Computer Graphics and Visualization Techniques
- Advanced Vision and Imaging
- Human Pose and Action Recognition
- Human Motion and Animation
- Medical Image Segmentation Techniques
- Image Retrieval and Classification Techniques
- Image Processing and 3D Reconstruction
- 3D Surveying and Cultural Heritage
- Robotics and Sensor-Based Localization
- Advanced Numerical Analysis Techniques
- Advanced Image and Video Retrieval Techniques
- Computational Geometry and Mesh Generation
- Optical measurement and interference techniques
- Data Stream Mining Techniques
- Physical Unclonable Functions (PUFs) and Hardware Security
- Graph Theory and Algorithms
- Advanced Authentication Protocols Security
- Remote Sensing and LiDAR Applications
- Machine Learning and ELM
- Data Visualization and Analytics
- Video Surveillance and Tracking Methods
- Face and Expression Recognition
- Image and Object Detection Techniques
- Artificial Immune Systems Applications
Chongqing University of Posts and Telecommunications
2022-2023
Chinese Academy of Sciences
2017-2019
Shenzhen Institutes of Advanced Technology
2017-2019
Shenzhen Academy of Robotics
2018
National University of Defense Technology
2005-2014
University of Science and Technology of China
2012
Cardiff University
2011
Three-dimensional surface registration transforms multiple three-dimensional data sets into the same coordinate system so as to align overlapping components of these sets. Recent surveys have covered different aspects either rigid or nonrigid registration, but seldom discuss them a whole. Our study serves two purposes: 1) To give comprehensive survey both types focusing on point clouds and meshes 2) provide better understanding from perspective fitting. Registration is closely related...
We perform co-analysis of a set man-made 3D objects to allow the creation novel instances derived from set. analyze at part level and treat anisotropic scales as shape style. The then allows style transfer synthesize new objects. key is correspondence, where major challenge handling large variations diverse geometric content in propose style-content separation means address this challenge. Specifically, we define correspondence-free signature for clustering. show that confining analysis...
We perform co-analysis of a set man-made 3D objects to allow the creation novel instances derived from set. analyze at part level and treat anisotropic scales as shape style. The then allows style transfer synthesize new objects. key is correspondence, where major challenge handling large variations diverse geometric content in propose style-content separation means address this challenge. Specifically, we define correspondence-free signature for clustering. show that confining analysis...
We present an algorithm for multi-scale partial intrinsic symmetry detection over 2D and 3D shapes, where the scale of a symmetric region is defined by distances between points region. To identify prominent regions which overlap vary in form scale, we decouple extraction performing two levels clustering. First, significant scales are identified clustering sample point pairs from input shape. Since different can share common point, shape covered clusters overlap. introduce matrix (SSM), each...
3D models of humans are commonly used within computer graphics and vision, so the ability to distinguish between body shapes is an important shape retrieval problem. We extend our recent paper which provided a benchmark for testing non-rigid algorithms on human models. This far stricter challenge than previous benchmarks. have added 145 new use as separate training set, in order standardise data provide fairer comparison. also included experiments with FAUST dataset scans. All participants...
We present a method for realtime reconstruction of an animating human body,which produces sequence deforming meshes representing given performance captured by single commodity depth camera. achieve single-view mesh completion enhancing the parameterized SCAPE model.Our method, which we call Realtime SCAPE, performs full-body without use markers.In estimations body shape parameters and pose parameters, needed reconstruction, are decoupled. Intrinsic is first precomputed subject, determining...
We have created a new benchmarking dataset for testing non-rigid 3D shape retrieval algorithms, one that is much more challenging than existing datasets. Our features exclusively human models, in variety of body shapes and poses. models humans are commonly used within computer graphics vision, so the ability to distinguish between an important problem. In this track nine groups submitted results total 22 different methods which been tested on our dataset.
Decision trees have been widely used for online learning classification. Many approaches usually need large data stream to finish decision induction, as show notable limitations (even fail) with small stream. In fact, there exist many real instances the paper, we propose a novel incremental extremely random forest algorithm, dealing classification streaming data. our method, arriving examples are stored at leaf nodes and determine when split combined Gini index, so can be expanded...
Feature matching is a challenging problem at the heart of numerous computer graphics and vision applications. We present SuperMatching algorithm for finding correspondences between two sets features. It does so by considering triples or higher order tuples points, going beyond pointwise pairwise approaches typically used. formulated using supersymmetric tensor representing an affinity metric that takes into account feature similarity geometric constraints features: cast as graph problem....
We perform co-analysis of a set man-made 3D objects to allow the creation novel instances derived from set. analyze at part level and treat anisotropic scales as shape style. The then allows style transfer synthesize new objects. key is correspondence, where major challenge handling large variations diverse geometric content in propose style-content separation means address this challenge. Specifically, we define correspondence-free signature for clustering. show that confining analysis...
PDF HTML阅读 XML下载 导出引用 引用提醒 支持在线学习的增量式极端随机森林分类器 DOI: 10.3724/SP.J.1001.2011.03827 作者: 作者单位: 作者简介: 通讯作者: 中图分类号: 基金项目: 国家自然科学基金(90707003, 60970094) Incremental Learning Extremely Random Forest Classifier for Online Learning* Author: Affiliation: Fund Project: 摘要 | 图/表 访问统计 参考文献 相似文献 引证文献 资源附件 文章评论 摘要:提出了一种增量式极端随机森林分类器(incremental extremely random forest,简称IERF),用于处理数据流,特别是小样本数据流的在线学习问题.IERF 算法中新到达的样本将被存储到相应的叶节点,并通过Gini 系数来确定是否对当前叶节点进行分裂扩展,在给定有限数量,甚至是少量样本的情况下,IERF 算法能够快速高效地完成分类器的增量构造.UCI...
We present a new benchmark for testing algorithms that create canonical forms use in non-rigid 3D shape retrieval. have combined two existing datasets to varied collection of models testing. Canonical attempt factor out shape's pose, giving pose-neutral shape. This opens up the possibility using methods originally designed rigid retrieval task demonstrate by it compare performance nine form methods, three different algorithms.
Abstract In this paper, we propose a fast and stable method for 2D shape deformation based on rigid square matching. Our utilizes uniform quadrangular control meshes shapes tries to maintain the rigidity of each in mesh during user manipulation. A matching is performed find an optimal pure rotational transformation mesh. An iterative solver proposed compute final result entire by minimizing difference between deformed vertices their counterparts neighboring square. The as possible details...
We present an implicit approach for pair-wise non-rigid registration of moving and deforming objects. Shapes interest are implicitly embedded in the 3D vector space. In this embedding space, is performed using a global-to-local framework. Firstly, non-linear optimization functional defined on distance function used to find global alignment between shapes. Secondly, incremental cubic B-spline free form deformation recover transformation parameters. Local posed terms minimising energy...