Snake-based approach for building extraction from high-resolution satellite images and height data in urban areas

摘要

In urban areas, it has been long a challenge topic to automatically extract urban objects from images duernto the high object density and scene complexity. Applying normal image processing methods could notrnachieve satisfied performance, especially for high-resolution satellite images. This paper presents arnsnake-based approach for building extraction from high-resolution IKONOS satellite images byrncombining height data. Multiple cues derived from both data sources are integrated into our snake modelrnto precisely control the movement of contours so that the boundary of buildings can be located with goodrnaccuracy and efficiency. Some experiment results are presented as well in this paper.