Kernel uncorrelated neighbourhood discriminative embedding for radar target recognition

摘要

A new manifold learning algorithm, called kernel uncorrelated neighbourhood discriminative embedding (KUNDE), is presented for radar target recognition. The purpose of KUNDE is to preserve the within-class neighbouring geometry, while maximising the between-class scatter. Optimising an objective function in a kernel feature space, nonlinear features are extracted. In addition, a simple uncorrelated constraint is introduced to get statistically uncorrelated features, which is desirable for many pattern analysis applications. Experimental results on both measured and simulated data demonstrate the effectiveness of the proposed method.