Automated street corridor classification using mobile mapping system

点击翻译按钮获取中文摘要

Olameter Corporation is a leading provider of utility services in North America, with its Canadian division based in Aurora, Ontario. This year, Olameter aims to start development on a new big spatial data acquisition and analytics service, which will be used for large-scale inventory management and rapid healthiness monitoring of key infrastructure, such as utility lines and road surfaces. To support this research, the company recently introduced IP-S3, the most compact high-definition mobile mapping system available today, which currently provides the company with 400 km of street corridor data daily. Feature extraction and information retrieval from such a massive volume of data generated by IP-S3 presents a high degree of processing complexity and requires a large amount of resources; however, this work is currently performed manually, which is a generally inefficient use of time and resources. In support of the development of Olameter's new big data service, the proposed research intends to investigate automated computer vision methods of efficiently classifying street scenes captured by the IP-S3 into key categorized objects, which will serve as a base information layer for further production of a vectorized infrastructure map.