Precision Terrain Modeling Approach in Complex Mountainous Areas Based on Compact UAV Ka-InSAR Data

Abstract

A high-precision digital elevation model (DEM) is useful for disaster investigation and evaluation in cloudy, rainy, and complex mountainous areas. However, clouds and rain make the optical images and laser point-cloud data acquisition difficult, while noise prohibits obtaining accurate surface information. Additionally, the complex elevation difference in mountainous areas increases the data processing difficulty, such as phase unwrapping (PU) and filtering. To overcome these problems, first, we introduce a new airborne multibaseline Ka-interferometric synthetic aperture radar (InSAR) system developed by the Beijing Institute of Radio Measurement. The system affords high resolution and small volume, is lightweight, has a good top-view angle, and is flexible. Thus, it reduces the flight platform’s dependence and improves the aircraft’s adaptability and universality. Moreover, a multibaseline PU method of a two-stage programming approach (TSPA) is selected to overcome the influence of severe noise and the phase continuity assumption limitation. Additionally, an adaptive filtering method for InSAR point clouds considering coherence and optimal bending energy is proposed. This method’s validity is verified using stereo satellite images, ground observation point precision checks, and geomorphic texture analysis against existing DEM results. The experimental results demonstrate that the proposed scheme has a good filtering effect on noise, vegetation, residential building areas, and bridges, significantly reducing manual intervention. Moreover, the results highlight that our method is well integrated with stereo images and has more texture details than conventional stereo mapping results, with a mean square error of elevation of 1.938 m.