Information Extraction of Urban Environments with Remotely Sensed Data

与其他自然景观相比，城市地区的遥感可能是最具挑战性的，因为土地利用和土地覆盖模式的复杂性，3D人造和自然特征，以及需要非常详细的信息。目前，世界上居住在城市地区的人口比居住在农村地区的人口多。获取城市地区的详细信息对于可持续城市发展和适应全球环境变化非常重要，例如海平面变化和对灾害的紧急反应。随着越来越多的航空和星载遥感数据以更高的空间分辨率记录不同波长范围的信息，现在可以有效地提取城市地区有用的地表信息。虽然遥感在许多城市中的应用已经取得了成功，但在遥感难以获取的信息的自动提取方面仍然存在一些具有挑战性的问题。为了准确、自动地提取地表信息，需要新的、改进的方法和算法对多传感器数据进行处理和集成。在这项提案中，我将重点填补利用遥感自动提取陆面信息方面的空白。具体地说，在这项建议中，我将开发基于多视点图像匹配的三维(3D)城市建筑物建模方法；使用航空图像的线性特征检测和分析提取城市道路表面状况信息，以及使用机载激光雷达估计表面粗糙度；以及通过在颜色信息中添加结构和形状信息来进行城市树种分类，以提高精度。在这一资助期间，将培训5名博士、6名硕士和额外的荣誉本科生。拟议的研究将为遥感知识贡献新的算法和思想，并有助于遥感产品在城市环境应用中的业务使用。城市规划者、市政府管理人员和决策者、城市交通主管部门、城市灾害管理团队、城市林业员、水文学家和生态学家将受益于3D建筑模型、道路状况评估和城市树种分类的研究成果。研究城市气候、水文、能源消耗和环境变化的研究人员可以将由此产生的信息作为他们的模型的输入。这项研究将在可持续城市发展、改善城市交通管理、提高加拿大人的生活质量、适应全球气候变化和自然灾害管理方面惠及加拿大。

Remote sensing of urban areas is probably the most challenging compared with other natural landscapes, because of the complexity of land use and land cover patterns, the 3D man-made and natural features and the need for very detailed information. Currently more people live in urban areas than rural areas in the world. Acquiring detailed information in urban areas is important for sustainable urban development and adaptation to global environmental change, for example sea level changes and emergency response to disasters. With the increasing availability of aerial and spaceborne remotely sensed data that record information from different wavelength ranges, with much higher spatial resolution, it is now possible to effectively extract useful land surface information in urban areas. Although remote sensing has been successful in many urban applications, there are still some challenging issues on automatic extraction of difficult to obtain information using remote sensing. New and improved methods and algorithms are required to process and integrate the multi-sensor data to extract the land surface information accurately and automatically. In this proposal, I will focus on filling the gaps in the automatic extraction of land surface information using remote sensing. Specifically, in this proposal, I will develop methodology for three dimensional (3D) urban building modelling based on multi-view image matching; urban road surface condition information extraction using linear feature detection and analysis from aerial images and surface roughness estimation using the airborne lidar; and urban tree species classification by adding structural and shape information to the colour information in order to improve the accuracy. During this grant funding period, five PhDs, six Master's and additional Honours undergraduate students will be trained. The proposed research will contribute new algorithms and ideas to the knowledge of remote sensing, and contribute towards the operational use of remote sensing products in urban environmental applications. Urban planners, municipal government managers and decision makers, urban transportation authorities, urban disaster management teams, urban foresters, hydrologists and ecologists will benefit from the research results of 3D building models, road condition evaluation and urban tree species classification. Researchers in urban climate, hydrology, energy consumption and environmental changes can use the resultant information as inputs to their models. The research will benefit Canada in terms of sustainable urban development, improving urban transportation management, improving the quality of life of Canadians, adapting to global climate change and natural disaster management.