Study On Scene Matching And Positioning Technology Of Missile-Borne SAR

As an active microwave imaging sensor,synthetic aperture radar(SAR)has been widely used in airborne and spaceborne platforms.With the continuous improvement of the requirements of the high-tech warfare for the anti-jamming and self-seeking ability of the seeker,a more precise guidance method is required in the final guidance stage to achieve a precise attack on high-value military targets.The scene matching guidance methods combined with SAR imaging technology can achieve the scene matching by using the SAR image,further calculate the position of the missile to correct the inertial navigation system error and the correction of the flight path deviation of the missile,which has become one of the most promising precision guidance method.This paper mainly focuses on the scene matching technology and the missile positioning technology in the scene matching guidance technology,which is divided into the following three aspects: 1.The basic principles and processes of SAR image matching and missile positioning technology,as well as the main technical difficulties are reviewed.First,the differences between missile-borne SAR images and optical images in noise characteristics and geometric distortion are analyzed.Subsequently,the inapplicability of optical image matching algorithm is pointed out.Then,the SAR gray image matching method based on normalized mutual information is introduced and simulated.Considering that the inertial navigation data need to be unified in different coordinate systems in the process of missile positioning,the conversion relationship between imaging coordinate system and geocentric coordinate system is first deduced,then the method of missile positioning based on single feature point is realized through simulation,and the conclusion that it is sensitive to error and insufficient positioning accuracy is obtained.2.Aiming at the inapplicability of traditional SIFT algorithm in SAR image matching and the low registration rate of traditional SURF algorithm,a SAR image matching method based on improved SIFT is proposed.First,the enhanced Gamma_MAP filter is used to suppress speckle noise,and the CLAHE method is used to enhance the SAR image,which will improve the contrast and highlight image feature.When assigning the main direction and constructing the feature descriptor,considering the problem that the traditional mean gradient operator is sensitive to noise,the ROEWA is used to instead of the mean ratio operator to calculate the pixel point gradient,which improves the robustness of the matching method to noise.In order to ensure the high registration rate of images,this paper adopts the two-step similarity measure strategy.The descriptor is first subjected to the RANSAC to obtain the rough matching pair of images.Furthermore,the bidirectional Euclidean distance between the feature vectors is calculated to match images.Finally,the simulation comparison among the above three matching methods and the simulation of measure date demonstrates that the SAR image matching method based on the improved SIFT has better performance.3.Aiming at the defect that the traditional slant range Doppler information based positioning method is sensitive to errors and cannot meet the guidance precision,a new method based on the improved Gaussian Newton-genetic hybrid algorithm is proposed.First,the latitude and longitude information of multi-feature points in SAR images after scene matching is obtained by using the positioning geometric model,whose origin point in the imaging coordinate system can be set as any point in the scene.After the coordinate transformation,the nonlinear equation is constructed by combining the corresponding slant range information.Then,the genetic algorithm and the Gauss-Newton method are combined to calculate the actual position of the missile in the imaging coordinate system because of their abilities in global optimization and local convergence.Finally,the latitude and longitude information of the missile in the geocentric coordinate system is obtained by coordinate inverse transformation,so that the trajectory deviation is corrected accurately.The feasibility and robustness of the proposed method are verified by error analysis and simulation experiments.