Search

Aiming at the limited space and resources of the UAV platform, the inaccurate target labeling and excessive calculation amount of high-resolution SAR image detection, a rapid detection method for UAV-borne high-resolution SAR image targets is proposed. Firstly, a bilateral filter is used to suppress the coherent speckle noise and non-uniform undulation inside the target in the high-resolution SAR image. Then, a large-scale SLIC segmentation is performed, and a small-scale SLIC segmentation is performed on the superpixel unit with under segmentation problem by analyzing the roughness of the superpixel unit. Finally, the over-segmented superpixel units are combined to obtain the target detection results. The experimental results show that this method can not only improve the detection efficiency of high-resolution SAR image targets, improve the SAR image target extraction ability, but also help to find small targets that are easily missed by some manual methods, and has good engineering application value.

Focus on the poor navigation precision when the aircraft suffers from the loss of satellite signals, the navigation method of the Synthetic Aperture Radar (SAR) platform based on Inertial Navigation System (INS), satellite positioning and scene matching is proposed. First, the corresponding equations of measurements are formed by the models of the INS, satellite positioning and scene matching points, respectively. And then applying the Kalman Filter (KF), the corresponding filter processes are established according to the different equations of measurements. The Extended Kalman Filter (EKF) is used to linearize the equations of satellite positioning and scene matching measurements. The proposed method can be executed when the satellites are not enough to provide the positioning information of the aircraft. In numeral simulations, when the satellite signals are partially missing, the proposed method can obtain lower absolute positioning errors than the INS and satellite-based method, and the proposed method obtains better precisian than the INS and satellite-based method when the satellite signals always exist. The simulation results prove the effectiveness and the superior performance of the proposed method.