Research On Information Fusion Technology Of Multi-source Satellite Observation For Ships

Ship surveillance via multi-source satellite observation information is an important development direction of space-based maritime surveillance.It can obtain more information on ships,and is of great significance in marine traffic management and the protection of maritime rights.Therefore,it is of great theoretical and practical value to study the information fusion technology of multi-source satellite observation for ships.This thesis mainly studies the target association,target tracking and other basic theories and key technologies in multi-source satellite observation information fusion to jointly improve the ship tracking and surveillance capabilities.The main contributions of this dissertation are as follows:1.Ship tracking via the Low Earth Orbit(LEO)electronic reconnaissance satellite network is studied.Aiming at the long and random revisit time of satellites,an Extended Kalman Filter(EKF)track filter algorithm based on mid-latitude voyage is proposed.The EKF is derived based on the mid-latitude voyage,and the effectiveness of the algorithm is verified by the Automatic Identification System(AIS)data and simulation.To track multiple ships and estimate the feature parameter of multiple emitters aboard using electronic intelligence satellites under the clutter interference,the long and random revisit time,and other complex conditions,a novel tracking algorithm using both kinematic(position and velocity)and feature information based on an improved Multiple Hypothesis Tracking(MHT)approach is proposed.A pre-processing of measurements is utilized via hierarchical clustering based on the position and characteristic of the emitter.The divergence information and fuzzy clustering are used to combine the feature information into the MHT framework to calculate the track score,and then the joint estimation of the target motion state and feature state is realized.Simulation results show that the proposed method has better tracking performance than the standard MHT algorithm.2.Ship tracking using the Geostationary Orbit(GEO)optical satellite is studied.Aiming at the large positioning error of the GEO optical satellite has due to the long distance from targets,the algorithms via multi-source information to improve the positioning accuracy are proposed.The static position correction is accomplished by the Rational Polynomial Coefficients(RPC)transform of coastline data,and template matching with the GEO image in image coordinates.The dynamic position correction is accomplished by the optimal association between AIS data and GEO satellite target points.Aiming at the low spatial resolution of GEO satellites,weak small ships,and many false targets due to the interference of clouds,islands and waves,an algorithm about ship detection and tracking is proposed.The Normalized Difference Water Index(NDWI)and block region statistics are used for image segmentation and land and cloud masking,and then the ocean regions are extracted.Detecting weak targets in complex sea backgrounds is solved by Constant False Alarm Rate(CFAR)processing and morphological filter.The MHT algorithm in geographic coordinates is used to track targets,further eliminate false targets,and the motion information of targets is obtained finally.The image sequence of Gao Fen-4(GF-4)and AIS data are used to verify the effectiveness of the algorithm,which shows its high engineering application value.3.Ship track association of the GEO optical satellite is studied.A Track Segment Association(TSA)algorithm based on the combination of motion information and amplitude information is proposed.The association strategy is that non-maneuvering targets are associated with motion information and maneuvering targets are associated with amplitude information aided.GF-4 satellite data is used to verify the effectiveness of the algorithm.Aiming at the synchronous track association between the GEO optical satellite and AIS under system error,an Iterative Closest point(ICP)algorithm based on multi-feature is proposed to correct the positioning error of GEO tracks using associated track.The effectiveness of the algorithm is verified by GF-4 satellite and AIS data.4.Ship association using multi-source imaging satellite is studied.Aiming at the joint ship surveillance with GEO and LEO imaging satellites,a multi-level target association algorithm is proposed.By using multi-feature information,such as position,heading and size,multi-level association is carried out,which improves the accuracy and robustness of target association in densely populated sea areas,and GF-4 and Gaofen-1(GF-1)satellite data is used to validate the effectiveness of the algorithm.Aiming at ship association using multi-source Synthetic Aperture Radar(SAR)satellite,a correlation algorithm based on position and size features and a method for velocity estimation by azimuth offset iteratively are proposed.The effectiveness of the algorithms is verified via GaoFen-3(GF-3)and Sentinel-1(S-1)satellite data.