Radar Target Recognition by MSD Algorithms on Angular-Diversity RCS

In pattern recognition, the maximum scatter difference (MSD) algorithm has physical significance similar to that of Fisher linear discriminant analysis (FLDA), i.e., looking for optimal projection-based features. The only difference is that the MSD adopts the scatter difference as discrimination criterion. Thus, the MSD will decrease the complexity of algorithm and then speed up calculation processes. It is usually applied to discrimination problems whose solutions cannot be directly obtained due to singularity of within-class scatter matrix. This letter implements target recognition by MSD algorithms on angular-diversity radar cross section (RCS). Numerical simulation shows that the MSD-based recognition scheme can not only accurately recognize unknown radar targets, but also have good ability to tolerate random fluctuations of environments.