Radar Adaptive Detection Algorithm Based On Clutter Structure

As an electromagnetic sensor,radar is widely used in the military field and the civil field.Target detection is the basis for subsequent operations such as position,recognition and track.Due to the complex background,it is difficult to collect sufficient target-free training data to meet the training requirement,when the traditional detection algorithms are applied to the practical scenarios.As a result,the accuracy of clutter covariance matrix estimation decreases,resulting in detection performance losses.For these problems,this thesis studies detection algorithms by exploiting the clutter covariance structures: in the small sample environment.It uses both persymmetry and spectral symmetry of the clutter.The clutter spectral symmetry means that it has a symmetric power spectral density and its covariance matrix is a real matrix.The main contributions of this thesis are summarized as follows:Firstly,clutter measured data were analyzed,including amplitude and structure of covariance matrix.The prior information provides an important basis for subsequent detection.In this thesis,a binary hypothesis test is utilized to design a method for detecting spectrum symmetry,which is verified by utilized simulated and measured data.Secondly,combining the spectrum symmetry and persymmetry of the covariance matrix,a new covariance estimation method is designed.Then,an adaptive matching filtering detection method based on the clutter covariance structures is proposed.Theoretically,under the same situations,the training sample data can be reduced to 1/4 without affecting the detection performance.The performance improvement in the target detection is verified by using both simulated and measured data.Thirdly,three distributed target detection algorithms(i.e.,the Rao test,Wald test and GLRT)are designed in compound-Gaussian clutter with gamma texture,by exploiting the persymmetric structure in the covariance matrix.All the proposed detectors ensure the constant false alarm rate property with respect to the covariance matrix structure.Numerical results based on both simulated and real data show that the proposed detectors outperforms the traditional detectors,when the secondary set is limited.In addition,the proposed Rao test is the most robust in the mismatched case.