室内移动三维测图点云数据的多元质量评价与修补研究

Mobile mapping system is one of the cutting-edge technologies to obtain three-dimensional information, and represents the mainstream development of geospatial data collection. The indoor scene geospatial data acquisition is the core application of mobile mapping system. Technologies and applications of outdoor mobile mapping system has become relatively mature. In terms of the unique situation of indoor environment as close-range, multi-occlusion, inconsistent illumination, and no absolute positioning information acquired etc., the indoor mobile mapping point cloud data suffers the problems of inaccuracy and incompleteness, which finally results in locally degraded point cloud data. The project innovatively proposes the multi-dimensional quality assessment for 3D indoor mobile mapping point cloud data, and explores the theory and method for degraded point cloud data repairing. The main contents of the project include: degraded elements modeling and degraded phenomena description for indoor mobile laser scanning point cloud data according to the diversity of the degraded elements and the locality of the degraded phenomena; multi-dimensional local quality assessment of non-reference point cloud data based on semi-supervised learning method and marked multi-dimensional quality point cloud dataset; local repairing of the degraded point cloud data on data layer using the good quality point cloud data based on compressed sensing theory. The outputs of the research will extend the theoretical system of point cloud data processing and benefit the applications such as indoor mobile mapping, digital city, culture heritage, etc.

移动测图系统是三维信息获取的前沿科技之一，代表地理空间数据采集的发展主流，其中室内场景是移动测图系统的核心应用。室外移动测图系统的技术和应用已较成熟。和室外场景相比，室内环境具有近距离、多转角、易遮挡、光照复杂、缺乏绝对定位等特点，室内移动三维测图获得的点云数据普遍存在多种降质，而且每一局部位置的降质情况都不同。本课题创新性提出，从室内移动三维测图点云数据的多元质量评价入手，探索点云数据修补的理论和方法。主要研究内容包括：研究室内移动三维测图点云数据多元降质要素的建模与降质现象的特征描述；引入半监督学习，结合多元质量标记数据集，研究无参考条件下点云数据的局部多元质量评价；在压缩感知理论框架下，结合多元质量评价结果，研究降质点云的局部修补方法。研究成果将拓展点云数据处理的理论体系，在移动测图、数字城市、数字文化遗产等方面具有重要意义。

移动测图系统是三维信息获取的前沿科技之一，代表地理空间数据采集的发展主流，其中室内场景是移动测图系统的核心应用。室外移动测图系统的技术和应用已较成熟。和室外场景相比，室内环境具有近距离、多转角、易遮挡、光照复杂、缺乏绝对定位等特点，室内移动三维测图获得的点云数据普遍存在多种降质，而且每一局部位置的降质情况都不同。本课题创新性提出，从室内移动三维测图点云数据的多元质量评价入手，探索点云数据修补的理论和方法。项目完成了以下研究内容：（1）设计多种平台的室内三维测图系统；（2）研究室内移动三维测图点云数据多元降质要素的建模与降质现象的特征描述；（3）引入半监督学习，结合多元质量标记数据集，研究无参考条件下点云数据的局部多元质量评价；（4）结合多元质量评价结果，研究降质点云的局部修补方法。研究成果将拓展点云数据处理的理论体系，在移动测图、数字城市、数字文化遗产等方面具有重要意义。