Feature extraction for identification of drugs and explosives concealed by body packing based on discrete cosine transform plus linear discriminant analysis

Abstract

In the analysis towards the energy dispersive X-ray diffraction spectra of drugs and explosives concealed by body packing, for solving the case of high dimension, small sample and poor signal-to-noise ratio, the method of feature extraction based on discrete cosine transform plus linear discriminant analysis (DCT plus LDA) is introduced. DCT is employed for dimensionality reduction and LDA is employed for feature extraction. The signal energy is concentrated into a few coefficients by DCT and LDA seeks directions efficient for discriminant. Principal component analysis (PCA) and DCT plus PCA were investigated to compare with the features extracted by DCT plus LDA. Then, in order to avoid the influence of classifiers, neural network (NN) and support vector machines (SVM) were introduced to classify the samples according to the features. It is shown that the recognition rates obtained by DCT plus LDA are more accurate (averaged recognition rate > 97.4%). This work has demonstrated that DCT plus LDA is effective in featured extraction for identification of drugs and explosives concealed by body packing.