Reconstruction of façades in point clouds from multi aspect oblique ALS
In this paper an approach is described which allows the reconstruction of windows in façades from airborne laser scanning (ALS) data taken in oblique view. A challenge is the sparse sampling of the façades by ALS compared to terrestrial laser scanning. In a first step windows are detected by a search for measurements of laser pulses which pass the windows (voyeur effect) and reflect behind the façade (indoor points). For a general study exploiting this side effect two different cases are considered: (i) a city model with faces of the façade is available and the point cloud has to be co-registered, (ii) a city model is not available and façade planes has to be automatically derived from the point cloud itself. Then, in both cases indoor points are mapped to a raster on the façade plane to accumulate hypotheses for the position of windows. Based on the points lying around these positions rectangular windows are reconstructed on the façade wall. Finally, the outlines of the windows are adapted by considering all windows of one façade. For a selection of eleven façades a quality analysis was performed showing a detection rate of 67%, a false alarm rate of 9% and an error of window sizes varying from a few centimeter to about one meter.