SVM and eigenspace projection methods are widely used in pattern recognition. A separating hyperplane and the projection axes are found for solving the two-class classification problem. The covariance matrices always represent the class scatters. In this
