Automatic Target Recognition Based on Discrete Wavelet Transform and Principal Component Analysis

This paper presents a method for automatic target recognition based on discrete wavelet transform (DWT). This proposed method is firstly detecting regions of interest in the original image and then selecting the three-level lowest frequency subimage of each region as the feature vector. To reinforce the performance, principle component analysis (PCA) is incorporated for dimensionality reduction of feature vectors. During classification, nearest feature space (NFS) classifier are presented for robust recognition in presence of varying viewpoints, illumination, and expression. Experiment results show that this proposed method achieves a significant improvement in automatic target recognition rate with high dimension reduction of the feature vector.