Dynamic Small Targets Detection Using Multi-temporal SAR Images

Synthetic Aperture Radar (SAR) imagery is widely used in many applications due to its independence from weather conditions. Target detection is an important application of SAR, aiming at automatic identification and localization of ground targets in images, such as buildings, vehicles, ships, etc. However, the clutter and background interference increase the difficulty in target detection, especially for small targets. In this paper, we propose a small target detection method based on multi-temporal SAR images. A dynamic frozen background is first generated from the multi-temporal SAR image data. Then, all test images are compared with the generated dynamic frozen background to get the changed candidates. Finally, a false alarming reduction is applied to remove wrong detection caused by the radiometric imbalance. Experiments based on real SAR data show that the proposed method can detect small targets with a low false alarming in a complex environment.