A comparative study on tea samples classification by E-tongue using cross correlation based PCA, PCA and Shamon´s Nonlinear mapping method

Present paper proposes the development of commercial tea sample authentication based on electronic tongue (E-tongue) instrumentation system using pulse voltametry method. The current signatures of E-tongue signals were processed by Principal component analysis (PCA) and Sammon´s nonlinear mapping to develop tea classifiers. Cross correlation coefficient between input and output signals of E-tongue have been also used for PCA analysis. The input voltage series has been used as a mask in enhancing pattern of all the tea classes. A relative comparison among three methods of feature extraction unambiguously establishes the highest distance of separation between unknown tea samples and cluster centroid of any of the six commercial tea samples.