A Comparative Study of Face Authentication Using Euclidean and Mahalanobis Distance Classification Method

In face recognition feature extraction and classification are the two aspects to be focused. In principle component analysis (PCA) based face recognition technique, the 2D face image matrices must be previously transformed in to one dimensional image vectors. In this paper two dimensional principle component analysis(2DPCA) is used to extract the features. Comparing to conventional principle component analysis, two dimensional principle component analysis is based on 2D matrices rather than 1D vectors. The image matrix is formed directly using original image matrices Recognition rate seems to be higher using two dimensional principle component analysis. The Mahalanobis distance is a metric which is better adapted than the usual Euclidean distance to settings involving non spherically symmetric distribution.Recall, precision, fmeasure, recognition rate are calculated and the results are analyzed for Oracle Research Laboratory (ORL) database and for the database taken using normal digital camera. This paper includes the comparison of Euclidean and Mahalanobis Distance classification methods and analyze the results. Keywordsfmeasure , precision, recall, recognition rate, Two dimensional principle component analysis(2DPCA)