Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection
IEEE Transactions on Pattern Analysis and Machine Intelligence
Unsupervised Feature Selection Applied to Content-Based Retrieval of Lung Images
IEEE Transactions on Pattern Analysis and Machine Intelligence
Two-Dimensional PCA: A New Approach to Appearance-Based Face Representation and Recognition
IEEE Transactions on Pattern Analysis and Machine Intelligence
Journal of Cognitive Neuroscience
Hi-index | 0.01 |
Linear dimensionality reduction (feature extraction) methods have been widely used in computer vision and pattern recognition. Two of the most representative methods are principal component analysis (PCA) and linear discriminant analysis (LDA). However, when dealing with a multidimensional dataset of dimension R^m^"^1@?R^m^"^1@?...@?R^m^"^N (e.g. for images N=2, videos N=3), these methods usually first transform the original data to high dimensional vectors in R^m^"^1^x^m^"^2^x^...^x^m^"^N, and then analyze the data in such a high dimensional space. In this paper, we propose a supervised dimensionality reduction method called neighborhood discriminative tensor mapping (NDTM), which can directly process the multidimensional data as tensors. Moreover, NDTM can make use of the local information of the dataset to achieve a better classification result. Experimental results on face recognition show the superiority of our algorithm to traditional dimensionality reduction methods.