SAR Image Simulation Based on SBR/PO Method for Polarimetric Feature Analysis

Due to high cost of the high-resolution polarimetric synthetic aperture radar (PolSAR) image, there are few literature on the automatic target recognition (ATR) of small target using polarimetric features. With the limitation of high-resolution POLSAR dataset, SAR image simulation can play an important role in the investigation of polarimetric features for man-made target. This paper presents a SAR image simulator and several polarimetric features extraction methods for small complex targets. The proposed SAR simulator is composed of a SAR echo simulator and an image formation algorithm. In order to generate the SAR echo efficiently and accurately, the shooting and bouncing rays (SBR) and physical optics (PO) hybrid method is employed for high frequency electromagnetic scattering prediction of complex objects. Based on the parameters of the moving and stationary target acquisition and recognition (MSTAR) dataset, the PolSAR images of military vehicles are created. In addition, the multi-reflection analysis can be done by decomposing the simulated SAR image into the images corresponding to the first, second, and third order reflections since SBR is a ray tracing based approach. After the simulated SAR image is generated, we employ target decomposition theorems to obtain polarimetric features for target interpretation. Using polarimetric decomposition, the scattering signals of the vehicle are decomposed into a weighted combination of various simple scattering mechanisms, which can be used to form a set of images representing different scattering features. In addition, a polarimetric feature map can be obtained by coloring the scattering centers according to their scattering mechanisms. The testing results show that the polarimetric features fit the multi-reflection analysis well for the simulated SAR image.