全自动深度相机三维扫描系统 - 计算机辅助设计与图形学学报.pdfVIP

第27 卷 第 11 期 计算机辅助设计与图形学学报 Vol. 27 No.11 2015 年 11 月 Journal of Computer-Aided Design Computer Graphics Nov. 2015 全自动深度相机三维扫描系统 杨红庄, 陆 炎, 方 清, 杨周旺, 刘利刚* ( 中国科学技术大学数学科学学院 合肥 230026) (lgliu@) 摘 要: 提出一个具有兼容性、完整性的3D 扫描自拍系统, 能够简单、快速地重建出带有丰富细节及纹理信息的人 体模型. 主要的特点是兼容现有的深度的扫描仪, 且建模重建过程只需通过简单的交互即可完成. 较之前的做法, 其 主要贡献是提出了一种基于距离场的模型补洞算法, 利用数据驱动方式获得较好的形状参数, 能够在不损失模型细 节的情况下, 自动补出头顶的形状, 避免深度相机无法扫描头发的缺陷, 使扫描变得更加方便, 快捷, 也提出了一种 针对深度相机的纹理自动贴图算法, 利用采集的深度数据驱动, 可自动计算纹理坐标, 均衡不同图片间光照的差别, 从而获得最佳的纹理效果. 关键词：三维扫描; 人体建模; 深度相机; 纹理贴图; 补洞 中图法分类号：TP391 Automatic 3D Scanning System via Depth Camera Yang Hongzhuang, Lu Yan, Fang Qing, Yang Zhouwang, and Liu Ligang* (School of Mathematical Sciences, University of Science and Technology of China, Hefei 230026) Abstract: This paper presents a 3D self-scanning system with both compatibility and integrity, which can quickly and easily reconstruct the human body model with rich detail and texture information. This system’s main features are compatible with the existing depth of the scanner, and the model reconstruction just to complete the simple interaction. Compared to the previous practice, the main contribution of this paper is to propose a algorithm about the filling the model hole which is based on distance field. Using the data-driven way, we get the a better shape parameter. This algorithm makes it possible to fill out the shape of the head automatically without losing model details, and to avoid the defects brought by depth camera without being capable of scanning hair, and thus to scan more convenient and faster. Also a new automatic texture mapping algorithm is proposed to deal with