Sparse representation-based feature extraction combined with support vector machine for sense-through-foliage target detection and recognition

Owing to multipath propagation effects of rough surfaces, scattering from trees and ground tend to overwhelm the weak backscattering of targets, which makes it more difficult for sense-through-foliage target detection and recognition. In this study, a novel method to detect and recognise targets obscured by foliage based on sparse representation (SR) and support vector machine (SVM) is proposed. SR theory is applied to analysing the components of received radar signals and sparse coefficients are used to describe target features, the dimension of the sparse coefficients is reduced using principal component analysis (PCA). Then, an improved SVM classifier is developed to perform target detection and recognition. A chaotic differential evolution optimisation approach using tent map is developed to determine the parameters of SVM. The experimental results indicate that the proposed approach is an effective method for sense-through-foliage target detection and recognition, which can achieve higher accuracy than that of the differential evolution-optimised SVM, SVM, k-nearest neighbour and BP neural network (BPNN).