检索结果-南通市图书馆

在线全文

学校读者我要写书评

暂无评论

基于混合结构的多视图三维场景重建

图学学报 2024年第1期45卷 199-208页

作者：周婧怡张栖桐冯结青浙江大学计算机辅助设计与图形系统全国重点实验室浙江杭州310058

基于PatchMatch的多视图高效和高精度三维重建一直是一个挑战性问题。红黑棋盘格模式的传播方式并行计算效率高,但对应的视图选择策略精度较差;基于马尔可夫链的视图选择策略能获取更为准确的匹配结果,但算法并行度较低。为了达成场景... 详细信息

基于PatchMatch的多视图高效和高精度三维重建一直是一个挑战性问题。红黑棋盘格模式的传播方式并行计算效率高,但对应的视图选择策略精度较差;基于马尔可夫链的视图选择策略能获取更为准确的匹配结果,但算法并行度较低。为了达成场景重建质量与重建时间的平衡,本文提出了一种基于混合结构的多视图三维重建算法,在第一阶段采用沿行/列并行的传播策略和马尔可夫链式的视图选择策略,得到质量较高的初始深度图,并通过引入多层次处理提升弱纹理区域的重建质量;在第二阶段采用棋盘格式传播方式和基于投票的视图选择策略提高计算效率,缩短重建所需时间。通过在Strecha和ETH3D数据集上进行了大量实验和对比表明,本文算法在不降低计算精度的前提下,计算效率提高2.5倍以上。

关键词：三维重建多视图立体匹配视图选择传播策略多层次处理

Automatic location and semantic labeling of landmarks on 3D human body models

在线全文

学校读者我要写书评

暂无评论

Computational Visual Media 2022年第4期8卷 553-570页

作者： Shan Luo Qitong Zhang jieqing feng State Key Lab of CAD&CG Zhejiang UniversityHangzhou 310058China

Landmarks on human body models are of great significance for applications such as digital anthropometry and clothing *** diversity of pose and shape of human body models and the semantic gap make landmarking a challenging *** paper,a learning-based method is proposed to locate landmarks on human body models by analyzing the relationship between geometric descriptors and semantic labels of landmarks.A shape alignmentalgorithm is proposed to align human body models to break symmetric ambiguity.A symmetry-awaredescriptor is proposed based on the structure of the human body models,which is robust to both pose and shape variations in human body *** regression algorithm is adopted to establish the correspondence between several descriptors and semantic labels of the *** and qualitative analyses and comparisons show that the proposed method can obtain more accurate landmarks and distinguish symmetrical landmarks ***,a dataset of landmarked human body models is also provided,containing 271 human body models collected from current human body datasets;each model has 17 landmarks labeled manually.

关键词： 3D human body model descriptor land-marking shape alignment

CNLPA-MVS:Coarse-Hypotheses Guided Non-Local PAtchMatch Multi-View Stereo

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Computer Science & Technology 2021年第3期36卷 572-587页

作者： Qitong Zhang Shan Luo Lei Wang jieqing feng State Key Laboratory of CAD&CG Zhejiang UniversityHangzhou 310058China

In multi-view stereo,unreliable matching in low-textured regions has a negative impact on the completeness of reconstructed *** the photometric consistency of low-textured regions is not discriminative under a local window,non-local information provided by the Markov Random Field(MRF)model can alleviate the matching ambiguity but is limited in continuous space with high computational *** to its sampling and propagation strategy,PatchMatch multi-view stereo methods have advantages in terms of optimizing the continuous labeling *** this paper,we propose a novel method to address this problem,namely the Coarse-Hypotheses Guided Non-Local PAtchMatch Multi-View Stereo(CNLPA-MVS),which takes the advantages of both MRF-based non-local methods and PatchMatch multi-view stereo and compensates for their defects ***,we combine dynamic programing(DP)and sequential propagation along scanlines in parallel to perform CNLPA-MVS,thereby obtaining the optimal depth and normal ***,we introduce coarse inference within a universal window provided by winner-takes-all to eliminate the stripe artifacts caused by DP and improve ***,we add a local consistency strategy based on the hypotheses of similar color pixels sharing approximate values into CNLPA-MVS for further improving ***-MVS was validated on public benchmarks and achieved state-of-the-art performance with high completeness.

关键词： 3D reconstruction multi-view stereo PatchMatch dynamic programming

Dual quaternion based virtual hand interaction modeling

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2013年第3期56卷 27-37页

作者： CHEN Wenfeng ZHU ShuJing WAN HuaGen feng jieqing State Key Laboratory of CAD & CG Zhejiang University

Virtual hands play key roles in three-dimensional interaction in various virtual reality ***,their visual realism has seldom been seriously discussed in the community of human computer interaction,not to mention the challenging balance between their visual/motion realism and real-time *** this paper,a novel approach for virtual hand interaction modeling based on dual quaternion is ***,dual quaternion blending(DQB) is introduced as pseudo muscular layer of our virtual hand model instead of traditional linear blending(LB) and quaternion blending(QB) due to its advantages in muscular deformation.A framework for virtual hand interaction is proposed which supports both gesture interactions and 3D *** results show that our proposed virtual hand interaction model is suitable for most virtual reality applications and achieves good visual/motion realism while maintains real-time performance.

关键词： virtual reality virtual hand interaction dual quaternion gesture interaction 3D manipulation

Copy-and-paste operation of planar polygonal shapes

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Progress in Natural Science:Materials International 2007年第7期17卷 859-866页

作者： Yang Wenwu feng jieqing Huang Shengsheng Jin Xiaogang(State Key Laboratory of CAD&CG,Zhejiang University,Hangzhou 310027,China) State Key Laboratory of CAD&CG Zhejiang University Hangzhou 310027 China

A 2D polygonal shape copy-and-paste method is proposed which is based on a multiple planar shapes blending algorithm. First sub-shapes are specified and selected,which correspond to user-defined visual features on the input *** they are copied and pasted with contribution weights to generate new shapes via a modified intrinsic 2D shape blending *** can edit the generated shape intuitively and interactively by adjusting the contribution *** proposed method fills the gap in the object modeling method- ology based on the copy-and-paste *** the static 2D copy-and-paste operation,the proposed method can also be applied to 2D metamorphosis among multiple planar shapes.

关键词： copy-and-paste 2D shape blending metamorphosis

Real-time rendering of algebraic B-spline surfaces via Bzier point insertion

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2014年第1期57卷 71-85页

作者： WEI FeiFei feng jieqing State Key Laboratory of CAD & CG Zhejiang University

This paper presents a GPU-based real-time raycasting algorithm for piecewise algebraic surfaces in terms of tensor product B-splines.3DDDA and depth peeling algorithms are employed to traverse the piecewise surface patches along each *** intersection between the ray and the patch is reduced to the root-finding problem of the univariate Bernstein *** polynomial is obtained via Chebyshev sampling points *** iterative and unconditionally convergent algorithm called B′ezier point insertion is proposed to find the roots of the univariate *** B′ezier point insertion is robust and suitable for the SIMD architecture of *** results show that the proposed root-finding algorithm performs better than other root-finding algorithms,such as B′ezier clipping and B-spline knot *** rendering algorithm can display thousands of piecewise algebraic patches of degrees 6–9 in real time and can achieve the semi-transparent rendering interactively.

关键词： real-time rendering algebraic B-spline surface root finding depth peeling

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

人体前景的自动抠图算法

计算机辅助设计与图形学学报 2020年第2期32卷 277-286页

作者：冉清冯结青浙江大学CAD&CG国家重点实验室

在基于立体视觉的人体建模系统中,背景像素的移除可以减少不必要的立体匹配计算,提高人体模型重建效率.为此,在给定大量具有前景Alpha蒙板真值的人体图像作为训练数据的前提下,提出了一个端到端的深度学习网络,以实现系统采集图像中人... 详细信息

在基于立体视觉的人体建模系统中,背景像素的移除可以减少不必要的立体匹配计算,提高人体模型重建效率.为此,在给定大量具有前景Alpha蒙板真值的人体图像作为训练数据的前提下,提出了一个端到端的深度学习网络,以实现系统采集图像中人体前景自动抠图.该深度学习网络包括2个阶段:人体前景分割阶段和人体前景Alpha抠图阶段.在人体前景分割阶段,采用Mask R-CNN网络中的目标检测和掩码生成2个负载,并结合训练数据进行迁移学习,得到了适用于人体前景二值化分割的模型网络.在人体前景Alpha抠图阶段,采用Encoder-Decoder网络架构实现Alpha蒙板的自动预测.首先引入核为5的非学习卷积层,以上一个阶段的二值化分割结果作为输入,自动得到三分图Trimap,再和人体前景训练数据一起作为此阶段抠图网络的输入;经过学习迭代,获得能够预测人体前景Alpha蒙板的模型网络.在实验部分,以单幅系统采集人体图像为输入,无需额外先验和人工交互,可以自动估计人体前景Alpha掩码结果.用户测试结果以及与其他方法的对比和分析证明了文中算法的可靠性和鲁棒性;同时,该自动抠图算法还对其他公开数据集的人体图像进行了掩码预测,实验结果表明该算法具有一定的泛化能力.

关键词：前景分割 Alpha抠图深度学习自动人体抠图

在线全文

学校读者我要写书评

暂无评论

表驱动的细分曲面GPU绘制

计算机辅助设计与图形学学报 2014年第10期26卷 1567-1575页

作者：黄韵岑冯结青浙江大学CAD&CG国家重点实验室杭州310058

为了充分利用GPU的并行计算能力高效地绘制递归定义的细分曲面,提出一种基于GPU的面分裂细分曲面的实时绘制算法.该算法通过离线预计算生成可以复用的细分查找表,它由细分矩阵组成,其大小仅与奇异点度数和最大细分深度线性相关,与输入... 详细信息

为了充分利用GPU的并行计算能力高效地绘制递归定义的细分曲面,提出一种基于GPU的面分裂细分曲面的实时绘制算法.该算法通过离线预计算生成可以复用的细分查找表,它由细分矩阵组成,其大小仅与奇异点度数和最大细分深度线性相关,与输入网格无关;对于细分曲面控制网格的每个曲面片,如果包含2个或2个以上奇异点,则进行一次局部预细分;之后对于不规则曲面片,利用细分查找表由初始控制网格直接计算得到各细分层次上的控制顶点,无需逐层计算,从而最大限度地发挥GPU的并行处理能力;最后对各层次上的规则曲面片使用硬件细分着色器绘制,大大提高绘制效率.实验结果表明,文中算法可以高效地绘制细分曲面的极限曲面.

关键词：表驱动不规则区域自适应 GPU实时绘制细分曲面

在线全文

学校读者我要写书评

暂无评论

视频内容敏感的动态字幕

计算机辅助设计与图形学学报 2011年第5期23卷 855-862页

作者：姜晓希冯靖怡冯结青浙江大学CAD&CG国家重点实验室杭州310058 浙江大学计算机科学与技术学院工业设计研究所杭州310027

针对传统方法中固定位置的字幕可能遮挡视频中观众感兴趣内容的问题,提出一种基于视频内容的动态字幕添加方法.通过综合考虑视频内容的重要性、视频画面之间以及相邻字幕间的连贯性来确定视频中视觉相对次要的区域,并在该区域添加字幕,... 详细信息

针对传统方法中固定位置的字幕可能遮挡视频中观众感兴趣内容的问题,提出一种基于视频内容的动态字幕添加方法.通过综合考虑视频内容的重要性、视频画面之间以及相邻字幕间的连贯性来确定视频中视觉相对次要的区域,并在该区域添加字幕,有效地减少了字幕对关键内容形成的遮挡;并提出了随视频内容动态确定字幕颜色以及通过对固定颜色描边2种方式,以增加字幕在视频画面上的对比度.实验结果表明,动态字幕可以有效地避免视频内容遮挡,提高视频播放效果.

关键词：视频内容能量函数色环动态字幕