Research Article

Fast Second-Order Orthogonal Tensor Subspace Analysis for Face Recognition

Algorithm 7

ODTSA.
Input: A set of sample matrices with class label information, , .
Output: left and right transformation matrices and .
(1) Initialize with an identity matrix;
(2) Until convergence Do:
   (2.1) Form the matrix ;
   (2.2) Form the matrix ;
   (2.3) Compute by solving the trace ratio optimization problem (27) with
      and ;
   (2.4) Form the matrix ;
   (2.5) Form the matrix ;
   (2.6) Compute by solving the trace ratio optimization problem (27) with
      and .
   End Do