Nearest neighbor classifier based on Riemannian metric in radar target recognition

A successful design for a nearest neighbor classifier based on Riemannian metric in radar target recognition is presented. In Riemannian space, obtaining feature coefficient using subspace methods can be regarded as an affine transformation, and the classifier can be deduced easily from the distance formula in Riemannian space. The classifier is compared with other classifiers and good performance is reported. This design for the classifier may serve as a guideline for dealing with the puzzle that how to combine feature extraction with classifiers reasonably in radar target recognition using range profiles.