基于细粒度理解的三维点云数据高质量建模及运动属性推理研究

高质量三维内容的生成是虚拟现实、智能制造、人机交互等行业领域的重要基础，已经成为了当前国际科学研究的前沿热点与难点，具有重要的研究意义和应用价值。随着消费级三维采集设备的快速普及，三维点云作为最常见的采集数据类型之一，已形成了海量的三维数据资源，且数据规模还在高速增长中。但由于采集的三维点云中包含大量噪声，质量偏低，导致其应用场景受到了极大的限制。同时，现有的三维数据大部分都是静态的，运动属性的缺失也是其无法广泛应用的另一个主要的原因。本项目针对高质量三维内容的广泛应用需求，开展基于三维点云的高质量建模与运动属性推理方法研究，将在三维CAD模型引导的三维点云细粒度解析、基于边缘检测的三维点云参数化曲线/曲面建模、基于细粒度部件的三维模型运动解析等方面取得理论和方法的实质性突破，为高质量三维内容的应用和推广提供理论基础和技术支撑，促进虚拟现实、智能制造、人机交互等领域的有机交叉和共同发展。

High quality 3D content generation are fundamental issues of virtual reality， intelligent manufacturing and human-computer interface. It has been widely studied and become the hottest research topics with great application values. With the rapid popularity of consumer 3D acquisition equipment, 3D point cloud, as one of the most common 3D data types, has formed a huge amount of 3D data, and the amount of data is still growing rapidly. However, due to the low quality of the collected 3D point cloud, its application scenarios are greatly limited. At the same time, most of the existing 3D data are static, and the lack of motion attributes is another main reason why it cannot be widely used. This project focuses on the scientific issues and application requirements of high-quality 3D content and is proposed to study high-quality modeling and motion attributes reasoning based on 3D point cloud. It will make technical breakthroughs on fine-grained analysis of 3D point clouds with 3D CAD model guidance, 3D point cloud parameterized curve/surface modeling, motion analysis of 3D models. We expect this project will bring theoretical basis and technological support for 3D high-quality content generation technique and software, facilitate the development of virtual reality, intelligent manufacturing, human-computer interface and interdisciplinary research, and facilitate the applications of high-quality modeling and motion attributes reasoning based on 3D point cloud technologies on research, security, education, culture, design, and entertainment.