Research On Target Recognition Technology Based On Simulation Model

Synthetic Aperture Radar(SAR)is an effective new system for observing the target radar.On the battlefield,the use of SAR images plays a very important role on automatic identification of the target.In this thesis,the simulation model is used to study the SAR image data generation and image target recognition.In this thesis,we pay attention on the SAR image simulation and how the simulation algorithms o f high resolution SAR images work in different background(such as terrain data with undulating characteristics,Buildings,ground typical vehicles,etc.),and finally,we do some relevant research about target recognition based on SAR images generated by the simulation algorithms.The main contents of each section are as follows:The first part,we mainly introduce two classifiers : the support vector machine(SVM)and K-nearest classifier,and then introduce a supervised feature extraction method for the target recognition of SAR images-Deep Brief Networks(DBN).Finally,based on the SAR image data generated by the simulation,the above feature extraction method and multiple classifiers are experimentally verified and analyzed.The second part,the main problem is the fast simulation of high resolution SAR images.In view of the widespread use of SAR image simulation based on echo signal need the original echo of the admission and an effective imaging algorithm,there is a great disadvantage in efficiency.Therefore,this thesis introduces the ray tracing algorithm into the SAR image simulation,and completes the transformation from the 3D model to the 2D SAR image according to the SAR imaging mechanism.However,the traditional ray tracing SAR image simulation has a lot of unnecessary computational overhead in the tracking process,which greatly affects the imaging efficiency.To solve the problem,this thesis introduces an adaptive scene partitioning method,The scene is divided into tree structure,so that in the tracking process can directly reduce the large number o f uncorrelated operations,that the simulation efficiency is greatly improved.The final experiment shows that the improved ray tracing algorithm has achieved considerable improvement in efficiency while ensuring the image quality.The thrid part is mainly aimed at the portability of the SAR image algorithm proposed in this paper under different application scenarios.Firstly,this algorithm is applied to the Digital Elevation Model(DEM)data that with ups and downs.In this thesis,the extended DEM data is triangulated by extending the original DEM data to match the resolution,and then the high resolution SAR image simulation of the terrain data is completed by means of the algorithm proposed in this paper.Secondly,The simulation of the ground ve hicle also proves the algorithm can be widely applicated.