A fast approach for dimensionality reduction with image data

作者：Benito M^*; Pena D

来源：Pattern Recognition, 2005, 38(12): 2400-2408.

DOI：10.1016/j.patcog.2005.03.022

• eigenfaces
• multivariate linear regression
• N-mode PCA
• principal component analysis
• singular value decomposition
• tensors

An important objective in image analysis is dimensionality reduction. The most often used data-exploratory technique with this objective is principal component analysis, which performs a singular value decomposition on a data matrix of vectorized images. When considering an array data or tensor instead of a matrix, the high-order generalization of PCA for computing principal components offers multiple ways to decompose tensors orthogonally. As an alternative, we propose a new method based on the projection of the images as matrices and show that it leads to a better reconstruction of images than previous approaches.

• 出版日期2005-12