A novel method of ISAR image pre-processing for ship

Abstract

Image pre-processing is the critical step in the ship target identification of ISAR images, which consists of speckle noise suppression and stripe noise suppression. In this paper, a pre-processing method of speckle noise and stripe noise suppression is proposed. For speckle noise estimation, an adaptive initial threshold is generated based on hypothesis of Rayleigh distribution, and the second segmentation threshold is calculated by the maximum gray value and gradient value in background region of ISAR image. For stripe noise estimation, a nonlinear compression method is used in the cross region of ship targets and stripe noise. In the experiments, the pre-processing results demonstrate the proposed method can not only suppress the speckle noise and stripe noise effectively, but also eliminate stronger noise points around the ship target.