Feature Enhancement Kernel Dictionary Learning for Radar Target Recognition

Radar target recognition technology is widely used in modern society. With the improvement of radar range resolution, radar echo target and jamming background echo can be distinguished by radar high-resolution information. This paper presents a feature enhancement kernel dictionary learning (FEK-DL) method for automatic radar target recognition (ATR). In order to enhance features of images for data enhancement and noise suppression, a matrix approximation method is used for multi-structure feature extraction. Instead of using traditional linear dictionary learning method, non-linear kernel function is used to map the targets into a high-dimensional space, in order to obtain a better classification performance. The training method and optimization steps of FEK-DL are presented in this paper. We carried out the experiment based on radar dataset to demonstrate the effectiveness of the proposed classification algorithm. The experimental results show that the classification algorithm has better classification performance than some representative dictionary learning algorithms.