Robust Moving Target Detection and Parameter Estimation Based on Three-channel SAR Images

Ground moving target indication (GMTI) is an important function of many airborne or spaceborne synthetic aperture radar (SAR) systems. The detection of ground moving targets based on SAR images, called SAR-GMTI, is a popular way. It is difficult for a conventional single-channel SAR system to achieve good GMTI performance, because of the difficulty in separating the ground moving targets from the clutter at low SCNR [1]. A dual-channel synthetic aperture radar along-track interferometry (ATI) has been used extensively to measure ocean surface currents [2]. In the oceanographic case, the entire ocean surface acts as a single, large target moving at a nearly uniform velocity. In the GMTI case, however, the objective is the detection of dim, discrete moving targets against a stationary scene. As is known, the performance of interferometric phase estimation suffers seriously from the accuracy of the image coregistration [3]. Therefore, it requires the coherent SAR images which are accurately coregistered. It is always difficult to coregister SAR images accurately with a long interferometric baseline. In addition, the ambiguity of the interference phase influences the performance parameter estimation. An appealing technology for GMTI is a three-channel SAR system. It is a very appropriate system that can achieve fine GMTI performance. In this paper, a novel robust approach for ground moving target indication (GMTI) is proposed aiming at a three-channel synthetic aperture radar (SAR) system and its real data. Throughout the following discussion, it is assumed that the SAR images achieved from different channels are coarsely coregistered using the cross correlation information of the SAR image intensity or other strategies after SAR imaging of the echoes acquires by each channel. Since the amplitude and phase errors from channel to channel will severely degrade the performance of parameter estimation, the adaptive channel equalization technology is used to balance or calibrate different channels. Array calibration operation should be done locally, since errors varies with different region, eigen-space decomposition approach can be used, due to the error information contained in the eigenvector corresponding to the maximum eigenvalue [4]. For a conventional SAR-GMTI system, SAR images are always required to be accurately coregistered. Otherwise, it will significantly influence the clutter suppression performance. It is true that it is difficult to reach a very good coregistration accuracy between images. In this paper, a robust method to image coregistration errors based on three-channel joint pixels data vector[3][5] is proposed to suppress clutter. The method can …