基于车载激光点云的城市道路三维精细重建

项目名称： 基于车载激光点云的城市道路三维精细重建

项目编号： No.41501493

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 天文学、地球科学

项目作者： 方莉娜

作者单位： 福州大学

项目金额： 20万元

中文摘要： 车载激光扫描系统在大规模城市场景信息获取以及城市道路三维精细重建中具有越来越明显的优势。目前车载激光点云的道路特征提取和三维重构存在智能化程度低、特征提取及几何重构的可靠性不足等技术瓶颈。本项目以智慧城市中重要组成要素--城市街景环境为研究对象，采用面向对象和机器学习的思想，将基于三维点云分类问题转化到基于几何目标识别问题，实现多结构、复杂城市场景的道路边界、交通标线等道路细节要素的感知和稳健提取；实现从数据驱动的道路目标三维表达到几何模型、语义知识驱动的道路多层次三维精细重构的转变。从而在最小人工干预条件下实现稳定、可靠的道路识别与建模，提高车载激光扫描数据的智能化解译程度，推进车载激光扫描系统在智慧城市、城市建模和城市管理等领域的应用。

中文关键词： 车载激光雷达；道路提取；三维重建；机器学习

英文摘要： Mobile laser scanning systems（MLS）have more advances in large-scale urban scene information acquisition and fine three-dimensional reconstruction of city roads. However, compared with advances in MLS, automated algorithms and software tools for efficiently and accurately extracting road features and reconstructing road geometric models rather fall behind. Using the theory of the object-oriented and machine learning, this proposal aims to extraction and delineate important elements of road—road boundaries and road markings from MLS point clouds based on geometric object recognition. This proposal will study the representation of urban road by fusing the geometric features and semantic description of road elements in order to reconstruct fine three-dimensional road models. The main contributions will provide a good solution for road extraction from MLS point clouds and give a promotion the wide use of Mobile laser scanning systems in many fields, such as smart city, 3D city modeling, and civil engineering.

英文关键词： Mobie laser scanning system;Road extraction ;3D reconstrcution;Machine learning