Digital image processing techniques to extract metric data on buildings from shadows on simulated air photos

Abstract

This paper investigates the use of shadows on aerial photographs to obtain ground parameters (length, width, height, area and volume) of buildings by image processing techniques. The assumptions of the flat ground and true vertical photography are made in the study, and the investigation is performed on a scale model of an urban environment. Only the case of separate shadows is examined, and no attempt is made to deal with the cases of overlapping shadows and shadows falling on other buildings. The numerical values of the building parameters obtained by image processing are compared with the corresponding “true” values measured directly on the scale model and a statistical error analysis is performed. The presence of a systematic error is detected in the height and volume residuals. A working model, based upon regression analysis, is suggested for use with actual aerial photographs.