The Role of Color Attributes and Similarity Grouping in 3-D Building Reconstruction

Abstract

This paper addresses two major issues: 3-D building reconstruction and the role of color attributes and similarity grouping. We present ARUBA, a general framework for automated 3-D building reconstruction from multiple color aerial images. After highlighting the strategy and concisely describing the framework and its 2-D and 3-D processing modules, we will evaluate the reconstructed roofs with respect to accurate reference data. The second part of the paper shows that geometry, although important, should not be the only source of information exploited in the reconstruction process. The main objectives are to demonstrate that (1) color is a very important cue in reconstructing a general class of objects, (2) it is crucial to retain all information during the entire processing chain, (3) a general class of objects parts can be efficiently extracted by grouping edges and lines by means of similarity, and (4) a mutual interaction between 2-D and 3-D processing is important.