Recent advances in detection and description of buildings from multiple aerial images

Abstract

A brief description of a method for detection and description of rectangular buildings from two or more registered aerial intensity images is provided. A new interactive editing module that can use partial results of the automated process to efficiently correct errors and derive complete descriptions is also described. The automated system operates by grouping features hierarchically to form roof hypotheses which are then verified by using wall and shadow evidence. Grouping and matching steps are interleaved and multiple descriptions are preserved when clear choices are not available. Some recent results are given.

This research was supported, in part, by the Defense Advanced Research Projects Agency of the United States Department of Defense under grant No. DACA76-97K-0001, monitored by the Topographic Engineering Center and in part by a subgrant from Purdue University under Army Research Office grant No. DAAH04-96-10444.