Face Recognition Based on Improved Tensor Neighborhood Preserving Embedding

Tensor Neighborhood Preserving Embedding(TNPPE) is a efficient subspace dimensionality reduction model for face recognition. However, the singularity problems is still remained. in the paper, we propose an Improved Tensor Neighborhood Preserving Embedding (ITNPE) algorithm for face recognition. We evaluate the algorithm by applying it to YaleB and AR databases. the experiments demonstrate excellent performance of our algorithm for the dimensionality reduction in face recognition.