A Multi-photo Least Squares Matching Algorithm for Urban Area Dem Refinement Using Breaklines

High quality digital elevation models (DEMs) are required for a large number of urban applications such as; telecommunication, urban development, and city planning and management. Sudden changes in surface topography weaken existing automatic terrain extraction techniques. The quality of such DEMs needs to be enhanced. This can be accomplished either manually or automatically. Manual enhancement is costly and time consuming. Hence, automatic enhancement is preferred. In this research, a new algorithm is presented to achieve high quality DEMs in urban areas. The proposed algorithm uses both breaklines and image signals in a least squares matching model. Errors in traditional least squares matching models occur due to incorrect modelling at breaklines when a continuous model is used in object surface reconstruction. Breaklines are utilized in this research to impose surface discontinuity in the least squares matching. In addition, the algorithm detect obscured pixels and eliminate them from the least squares matching model. The algorithm is tested using one meter resolution DEMs. Results for five test sites showed that the RMS are reduced from ten to five centimetres. This supports the use of the presented algorithm to refine DEMs in urban areas.