Change detection of concealed targets using repeat-pass SAR images

This paper proposes a novel change detection algorithm for locating target concealed by foliage using multi-pass low-frequency SAR images. Having registered the image pair, an intensity ratio detector is used to detect the potential changes due to disturbance in local mean backscatters with an computed threshold according to a given probability of false alarm. Once a local change is reported, CFAR detection is used to confirm it. The background estimation is operated on one of the local patch pairs according to the change style. Finally, the algorithm is tested on the benchmark image pairs from CARABAS-II data collected during the flight campaign in Vidsel 2002, which has show good performance in detection the arrival targets and departure targets while keep a rather low false alarm rate. The algorithm is also less time consuming.