Robust Height Reconstruction of Buildings Based on Esprit-Tomosar

Abstract

TomoSAR is one of the methods which can perform three-dimensional (3-D) imaging with multiple SAR datasets by using Direction-of-Arrival (DOA) estimation techniques to reveal height distribution of scatterers. However, resolution of the TomoSAR is generally insufficient for height estimation of small objects on the ground surface. Also, if we employ a high resolution DOA estimation method for the height estimation, its performance is often degraded due to orbital error. To solve this problem, we employ ESPRIT algorithm and propose a new method named ESPRIT-TomoSAR. The ESPRIT algorithm is one of high resolution DOA estimation techniques using orthogonality between signal and noise subspace, it has not only high resolution performance equivalent to the MUSIC algorithm but also robustness for orbital error. In this paper, to verify effectiveness of the proposed method, we provide experimental results by using actual airborne SAR dataset. The experimental results show that the proposed method realizes high resolution and robustness for orbital error, and it is effective for a precise height estimation such as the structure analysis of buildings.