Generation of large scale 3-D city models using InSAR and optical data

Abstract

Interferometric synthetic aperture radar (InSAR) techniquesare powerful tool for reconstructing the 3-D position of scat-terers, especially for the urban areas. Since the estimationaccuracy depends on the inverse of number of interferogramsand signal-to-noise ratio (SNR), it is necessary to use as manyas possible interferograms in order to achieve more accurateresult. However, the number of interferograms of TanDEM-Xdata is generally limited for most areas. Therefore, in orderto maintain the estimation accuracy, one feasible way is toincrease the SNR. In this work, we proposed a novel frame-work, which integrates the non-local procedure into SARtomography inversion and combines the robust estimation.A large-scale demonstration has been carried out with fiveTanDEM-X bistatic data, which covers the entire city of Mu-nich, Germany. Quantitative evaluation of the reconstructedresult with the LiDAR reference exhibits the standard devi-ation of the height difference is within two meters, whichimplies the proposed framework has great potential for highquality large-scale 3-D urban modeling.