Extraction of Rectangular Boundaries from Aerial Image Data

A novel approach for the extraction of rectangular boundaries from aerial image data is proposed and presented in this paper. In this approach, a centroid neural network (CNN) with a metric of line segments is also proposed for connecting low-level linear structures or grouping similar objects. Extracting rectangular boundaries for building rooftops from an edge image without height information of buildings such as stereo pairs or digital elevation models is very challenging and difficult. We introduce an approach that involves a combination of many algorithms to identify rectangular boundaries in a bottom-up manner. After the straight lines are extracted from an edge image using the CNN, rectangular boundaries are then found by using an edge-based grouping approach originating from perceptual concepts. We present the steps of the proposed building rooftop recognition method and experimental results on an edge image.