Research On The Acceleration Of Infrared Radiometric Signature Attitude And Shape Inversion For Space Targets

Space target recognition based on infrared imaging has extensive application in space awareness and midssile homing.Usually,targets are too small or distant to lend themselves to resolved imaging,so the acquired image contains littile information about shape and texture,which makes recognition a difficult problem.Exploiting temporal infrared radiometic measurements of target for shape and attitude estimation,i.e.infrared radiometric signature inversion,serves as an important technique to extract useful features for recognition.However,performing a complete signature inversion analysis requires extensive computation,preventing its application on space-based and warhead-based platform.So improving the efficiency of infrared signature inversion is of great value,which serves as the motivation of this thesis.Focusing on space warhead targets,the major contributions of thesis are as follows.Simulating the infrared signature generates the data for inversion research in which the precsion is very important.For the simulation of attitude motion for non-regular precession,the accumulation of attitude error caused by long time numerical integration has been a serious problem.In this thesis,based on the periodicity of ellipsoid function,a new attitude computation method is proposed which limits integration time in half rotation period.Attitude motion model is related to the inertial characteristics and initial attitude of space target,servers as important discriminant features.This thesis try to discriminate attitude motion mode based on the temporal variation and the frequency spectrum of the infrared radiometric signature based on simulating data.Regular and non-regular SAM(short axis mode)and LAM(long axis mode)of space target are characterized with the Euler angles and the temporal variation and frequency spectrum of simulated signatures generated under different attitude motion modes are analysed.The results show that infrared signature helps discriminate between rotation and precession,but is not useful in discriminate between sub-modes of precession.The computing complexity of the radiometric signature inversion based on facets radiometric model is reduced by simpily the shape model.One infrared radiometric signature inversion algorithm to estimate half-cone angle and nutation angle of conical targets is proposed.The mathematical expression for the projection function is constructed and then the infrared signature model is constructed.To address the problem of surface temperature variation,the sliding window estimator is used to estimate the half cone angle and nutation angle.In comparison with the algorithm based on the facetes model,the proposed method uses much simpler shape model,reduces the computation,and improves attitude and shape inversion precision when the temperature of object surface is varing.When addressing targets with shape of cone-cylinder,cone-ball,etc.,the cone model cannot be used to represent their shape,the inversion method based on cone model cannot be applied.Design new shape model with sufficient repsent power for these targets but with low complexity is critical for efficient inversion.To address this problem,a new shape model for axis-symmetric body is proposed,in which the surface of axial-symmetric body is approximated by a couple of trunctated concial surfaces and the projection of axial-symmetric body is approximated by the linear combination of a couple of base projection functions.The complexity of shape model is much lower in comparion with facets model.The least square estimator is used to extract the angular momentum direction,nutation angle and shape representation and the optimization is improved by separating the attitude and shape parameters.Simulated experiments show that the algorithm is effective in estimating the angular momentum direction and nutation angle,but the shape representation is difficult to estimate.Esimating shape and attitude parameters at the same time is the problem faced by all the inversion method based the shape model,this thesis try to use the relative motion between the target and sensor for shape-independent attitude analysis.Based on the analysis of synodic frequency model for light curve,the synodic frequency model of infrared signature for space target is constructed.Fouier series method is used to estimate the synodic frequency of infrared radiometric signature and then least square estimator is used to estimate the angular momentum direction and the precession frequency.In comparison with the method based on shape model,this scheme avoides the spade modeling and the optimization computatiom on shape parameters,reduces the computation complexity dramatically.Simulated expreiments show that the synodic frequency model of infrared signature is effective and the paramters can be estimated based on the relative motion information.