A divergence mean-based geometric detector with a pre-processing procedure

This paper proposes a divergence mean-based geometric detector to deal with the problem of target detection in a clutter with the limited sample data. In particular, a covariance matrix is used to model the correlation of sample data in each cell in one coherent processing interval. This modeling method can avoid the poor Doppler resolution as well as the energy spread of the Doppler filter banks result from the fast Fourier transform. Moreover, a pre-processing procedure, conceived from the philosophy of the bilateral filtering in image denoising, is proposed and combined within the geometric detection framework. As the pre-processing procedure acts as the clutter suppression, the performance of geometric detector is improved. Numerical experiments and real clutter data are given to validate the effectiveness of our proposed method.