Feature Extraction of 2D Radar Profile via Double-Sides 2DPCA for Target Recognition

Target recognition based on 2D radar profile is a rising application of radar technology, because 2D radar profile can offer more structural information about targets. But the methods of feature extraction based on 2D radar profile are very scarce. In this paper, a new approach double-sides 2DPCA is presented. It is performed by using original image matrices directly, while PCA always needs image matrices to be transformed into 1D vector. Besides, it can reduce the dimension of image from two sides and obtain a feature image with a much smaller size, while traditional 2DPCA, mainly used in the area of face recognition, can only reduce the size of image from one side. This approach has excellent dimension reduction property. The experimental results show that double-sides 2DPCA is effective on feature extraction of 2D radar profile and in much smaller dimension level, it has better recognition performance than traditional 2DPCA.