RESEARCH ARTICLE

Segmentation-aided phase unwrapping for 3D reconstruction with airborne array-InSAR images

The Key Laboratory for Information Science of Electromagnetic Waves (MoE), Fudan University, Shanghai 200433, China

^* Corresponding authors (emails: fm_hu@fudan.edu.cn (Fengming Hu); fengxu@fudan.edu.cn (Feng Xu))

Received: 11 December 2023
Revised: 25 March 2024
Accepted: 26 April 2024

Abstract

Unmanned aerial vehicle (UAV) array InSAR is a new type of single-flight 3D SAR imaging system with the advantages of high coherence and resolution. However, due to the low altitude of the platform, the elevation ambiguity of the system is smaller than the maximal terrain elevation. Since the spatial phase unwrapping is conducted based on the assumption of phase continuity, the inappropriate ambiguity will cause significant unwrapping errors. In this work, a 3D phase unwrapping algorithm assisted by image segmentation is proposed to address this issue. The Markov random field (MRF) is utilized for image segmentation. The optimal spatial phase unwrapping network is achieved based on the segmentation results. The asymptotic 3D phase unwrapping is adopted to get the refined 3D reconstruction. Results based on the real airborne array-InSAR data show that the proposed method effectively improves the elevation ambiguity.