Improved kernel fisher nonlinear discriminant analysis used in face identification

Local linear embedding proposed that face data would found in some low dimensional subspace, All face data would be linearity denoted optimal with data in neighborhood of the data. The input space without linear separability be mapped into linear divisible high dimensional space by nonlinear map-ping. Structure kernel spread inner matrix based on local linear embedding and kernel fisher nonlinear discriminant analysis, The matrix is nearly full rank. Make the optimal eigenvector in nonnull subspace of this matrix to test, and make a compare to kernel fisher null space algorithm. The experiment show the new algorithm is effective.