Robust Line Detection of Synthetic Aperture Radar Images Based on Vector Radon Transformation

Abstract

Line detection on synthetic aperture radar (SAR) images is still challenging due to the existence of strong speckles. This article proposes a novel line detection method for SAR images based on vector Radon transform (VRT). First, the ratio of exponentially weighted averages operator is determined to calculate the edge map. Then, the VRT is proposed to transform the edge map into the parameter map and decompose the parameter map along the angle of projection into two mutually perpendicular components, i.e., parallel component and vertical component, which strips off part of the random edge information and ensures the sharpness of the peaks (local extremum). Peaks are detected in the parallel component. Finally, the intersection analysis technique is adopted to remove the redundant lines. Experiments are carried out with six amplitude-format SAR images of various bands, resolutions, and polarizations from Chinese airborne SAR systems, GaoFen-3 (GF-3) satellite, and TerraSAR-X satellite. The results demonstrate the outperformance of the proposed method. Furthermore, a road edge detection scheme is designed based on the proposed line detection method. The experimental results of the road edge detection further confirm the effectiveness of the proposed method and show its potential for practical use.