Scattering Center Extraction Algorithm And Application Based On Three-Dimensional Imaging

In today's society,the electromagnetic scattering phenomenon of optical zone radar is widely used in the field of target recognition and feature acquisition.The accurate and effective scattering center model can effectively describe this phenomenon.Therefore,in the current electromagnetic field,the scattering center model is the basis of target characteristic analysis and recognition.There are currently two methods to model the scattering center: the forward method and the inverse method.Due to instability and subjectivity,the positive method is not widely used in the center modeling.At present,the inverse method is mainly applied in the modeling and analysis of the scattering center.It,also known as scattering center reconstruction,is divided into two implementation modes: “small angle model reconstruction” and “full attitude model reconstruction”.However,the currently used inverse method has disadvantages such as the need for strong calculation example support and the high complexity of the selection of the scattering center position.Based on the respective deficiencies of the above modeling methods,this article attempted to propose a new algorithm to overcome the weakness points of the current approaches.Firstly,this article combined the three-dimensional imaging algorithm with the machine learning algorithm,so that the user could select the number of scattering centers from two deci-sions: fixed compression ratio and fixed distance between the scattering points.This innovation made the selection process of scattering points more flexible,intelligent and efficient,and also made the position information of the scattering points more geometrically and physically mean-ingful.Secondly,based on the solution of linear system and GTD model in numerical analysis,this paper proposed a scattering center extraction algorithm based on three-dimensional imaging.This innovation of this paper formulated the extraction process of the amplitude information of the scattering points in the scattering model,reduced the computing power required in the process of extracting the scattering center,and explored the possibility of using the scattering center model extracted by this algorithm to interpolate the scene application.Finally,this paper provided different optimization strategies for the two dimensions of up-stream data source and training process for the general problem of “scattering center extraction algorithm based on three-dimensional imaging”.This paper then proposes an adaptive train-ing angle segmentation algorithm based on dichotomy and a reference value given algorithm based on greedy ideas,which effectively increases the robustness of the current algorithm and its applicability in different scenarios.This article not only gives the performance loss after applying the optimization strategy based on the calculation example for readers' reference,but also explores the full posture model reconstruction with the optimization strategy proposed in the process of “small-angle model reconstruction”.