A New Method Based on MDA to Enhance the Face Recognition Performance

A novel tensor based method is prepared to solve the supervised dimensionality reductionproblem. In this paper a multilinear principal component analysis (MPCA) is utilized to reduce thetensor object dimension then a multilinear discriminant analysis (MDA), is applied to find the bestsubspaces. Because the number of possible subspace dimensions for any kind of tensor objectsis extremely high, so testing all of them for finding the best one is not feasible. So this paper alsopresented a method to solve that problem, the main criterion of algorithm is not similar toSequential mode truncation (SMT) and full projection is used to initialize the iterative solution andfind the best dimension for MDA. This paper is saving the extra times that we should spend tofind the best dimension. So the execution time will be decreasing so much. It should be noted thatboth of the algorithms work with tensor objects with the same order so the structure of the objectshas been never broken. Therefore the performance of this method is getting better. Theadvantage of these algorithms is avoiding the curse of dimensionality and having a betterperformance in the cases with small sample sizes. Finally, some experiments on ORL andCMPU-PIE databases are provided.