作者: Chungan Lin , Andres Huertas , Ramakant Nevatia
DOI: 10.1007/978-3-0348-9242-1_13
关键词:
摘要: We describe a system for detection and description of buildings in aerial scenes. This is difficult task as the images contain variety objects. Low-level segmentation processes give highly fragmented segments due to number reasons. use perceptual grouping approach collect these fragments discard those that come from other sources. shape properties this. shadows walls help form verify hypotheses generated by process. latter step also provides 3-D descriptions buildings. Our has been tested on examples able work with overhead or oblique views.