Building extraction from LiDAR and aerial images and its accuracy evaluation

摘要

There are numerous building footprint extraction methods from various data sources and extraction strategies. However, the lack of a consistent standard to evaluate the accuracy of such methods impedes the comparison of different methods, as well as further exploration of new methods. In this study, a comprehensive evaluation system for building detection is proposed and implemented. In order to test its performance, four different building footprint extraction methods have been designed using LiDAR and Colour Infrared Aerial Images. The comparison of the proposed accuracy evaluation indices and the analysis of the results demonstrate that the designed evaluation system can provide a more complete assessment for building footprint detection.