Ground moving target indication in a SAR image based on background cognition

Ground moving target indication (GMTI) with synthetic aperture radar (SAR) is a very hot research topic in recent years. The traditional methods are based on multi-antennae technology, such as displaced-phase-center-antenna [1], along-track interferometry [2], and space time adaptive processing [3], and many space-borne or air-borne radars have been developed and put in use. However, the multi-antennae based technology is very complex and costs so much. Many researchers try to perform the tasks by using a single-antenna SAR. For example, J. R. Fienup detects the moving target by using autofocus technology based on shear averaging method [4], J. Dias et al. use the antenna radiation pattern information to indicate the moving targets and estimate their velocities [5], and G. Lv et al. detects and estimates the azimuth velocity component based on the symmetric defocusing method [6]. They can give fairly effective results. However, the mentioned methods exploit the phase difference between the background and moving targets, and the effectiveness will be influenced by clutters and interference. A new GMTI scheme used to detect moving targets with range velocity components is proposed in this paper based on background cognition. The scheme consists of four parts: a normalized background Doppler spectrum set (BDSS), a correlator used to extract the background of a given azimuth vector in a SAR imagery, a normalized moving targets´ Doppler spectrum set (MTDSS), and a classifier used to optimally classify the moving targets according to the waveform design algorithms [7]. The output of the classifier are feed back to the correlator to estimate the background further. As a result, the classifier gives more accurate estimations of moving targets. It is a closed-loop procedure. The scheme works well in our recent experiments.