ISAR Target Recognition Based on Capsule Net

Inverse synthetic aperture radar (ISAR) has been widely applied in automatic recognition of sea, air, and space targets thanks to its all-day, all-weather, and long-range operational abilities. But because of the inherent distinct characteristics such as the time-varying unknown effective rotation vector and the changing range and azimuth sampling rates, the field of ISAR ATR is facing many challenging problems such as image rotation-sensitivity, scaling-sensitivity, and translation-sensitivity, etc. This paper proposes a robust recognition method based on Capsule Net. In this structure, Capsule Net can extract the robust features of the deformable ISAR image. For the ISAR image database generated from electromagnetic (EM) echoes of four satellites, the proposed structure achieves higher recognition accuracy than traditional methods.