An Artificial Neural Network to improve orthogonal Laplacian face recognition.

摘要

In appearance based face recognition, choice of an algorithm for face recognition is based on the algorithm's ability to find the characteristics of an image manifold. The Orthogonal Laplacianface applies the Orthogonal Locality Preserving Projection (OLPP) algorithm to face recognition problem and creates a categorical and characterized map of its image dataset. Since the Orthogonal Laplacianface has a strong locality preserving power based on its orthogonal condition, there have been some reported challenges building geometrical map from its face sub manifold. In this thesis, an Artificial Neural Network is developed to address the existing problems and search for a solution to improve the overall face recognition process. Artificial Neural Networks are suitable for finding a solution for highly nonlinear dynamic system models.