Search

In face of the feature extraction of face recognition, this paper proposed a new method of local neighborhood based multi-manifold metric learning for face recognition.For small sample size problem in face recognition, it pre-processed the face image by using the feature face and used the Euclidean distance to select the nearest neighbors of each data point in the pre-processed face data set. It obtained the local weight matrix by this and the error distance between the reconstructed data points and calculated the original data points.At the same time,using the image set modeling manifold, it used the affine hull to represent the dataset information of manifolds and calculated the distance metric matrix among manifolds.By maximizing the distance between the manifolds and minimizing the distance between the data points and the reconstructed data points to find the dimension matrix. A large number of comparative experiments on the face data set show that the method is accurate and effective. [ABSTRACT FROM AUTHOR]