Application of computer vision techniques in modern SAR systems

Abstract

Synthetic aperture radar is a popular instrument for high-resolution imaging. Usage of state-of-the-art image and signal processing solutions allows to significantly increase the efficiency of such systems. In the paper two important applications of computer vision algorithms are described. In particular, usage of local feature extraction algorithms for the radar image stitching. This results in the automatical panorama creation without usage of the navigation data. In addition, the road location approach based on the stroke width transform and contour analysis is proposed. Application of the developed methods is illustrated with real SAR data examples.