Adaptive Manifold Discriminative Regression for Multi-view Radar HRRP Target Recognition

High-resolution range profiles (HRRP) is an important tool for automatic recognition of targets in radar systems. However, single-view HRRP is sensitive to the angle of aspect and provides limited information of target scattering feature, leading to unsatisfying recognition accuracy in practice. Multi-view HRRP from multiple radars or multiple scans appears promising for alleviating these problems and promoting recognition performance. In this paper, a novel adaptive discriminative regression algorithm based on manifold learning is advanced for multi-view radar HRRP target recognition. By using the adaptive locality preserving projection, our method can explore the geometric distribution and angular patterns of the data. Then, a novel cost based adaptive weighted multi-view least squares regression is proposed with the l<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2,1</inf> norm regularization to facilitate multi-view fusion and enhance feature selection capabilities. Experimental outcomes demonstrate that the presented algorithm remarkably enhances target recognition accuracy compared with existing methods.