Research On Typical Ground Target Detection And Recognition Method Based On Missile-borne SAR Imaging

The high-resolution real-time image acquired by the missile-borne SAR can not only correct the errors of inertial navigation system(INS)to control the missile hit the target accurately,but also carry out image detection and recognition,which has great application value for subsequent target location and even tracking.According to the motion characteristics of the missile in the flat-flying section,the key technologies of missile-borne high squint imaging and SAR image target detection and recognition are analyzed.The main works are as follows:1)The large squint imaging of missile-borne SAR in the flat-flying section is studied.For the serious coupling of range and azimuth of Missile-borne SAR in high squint mode,the traditional imaging algorithm is easy to cause image defocusing.An improved CS imaging algorithm based on time-domain correction of range walk is used.Firstly,the coupling is reduced in the time domain by range walk correction,and then the equivalent CS imaging algorithm is used for subsequent analysis.Third phase compensation is introduced in this paper to improve the performance of the algorithm.The effectiveness of the proposed algorithm for large squint imaging is verified by point target simulation.2)Target detection in SAR images is studied.Aiming at the coherent imaging characteristics of SAR,the target image is easily drowned by noise,and the extracted geometric structure features depend on the accuracy of target segmentation,this paper achieves target detection by fusing SAR image geometric structure and strong scattering point features.Considering the difficulty of SAR image acquisition,the scattering center model of the target is established through simulation to obtain the strong scattering point features,and then the structure features,contour features and strong scattering point features extracted from the model are combined for target detection.3)Target recognition in SAR images is studied.With the advantage of machine learning,the essential features of targets can be extracted step by step.This paper adopts the method of SAR image target recognition based on KNN.The effectiveness of the algorithm is verified by MSTAR public data set.The simulation results show that,the recognition rate of ground objects such as armored vehicles and tanks can basically reach 98% in a general condition.4)A software system for Missile-borne SAR imaging and target detection and recognition is designed by programming.