Automatic Building Outline Reconstruction Using 2D Building Data and Stereo Images

In this paper, we proposed an automatic procedure to reconstruct building outline using building data and stereo aerial images. Our methods focused on the outline of building, excluding the inner structure of roof surface. This procedure includes five steps, 1) to produce the edge gradient images by Canny Detector, 2) to defined the valid workspace in object space and image space by building data, 3) to detect line segments using grayscale Hough Transform, 4) to obtain 3D line candidates under epipolar geometric constraint and in valid workspace, 5) to reconstr uct the building outline from candidates by shortest route decision. The reconstruction results of six buildings with plane roof and nine buildings with gable roof showed that this procedure was quite reasonable.