(2) Principal Component Analysis (PCA)
Principal component analysis (PCA) is mainly used for data compression and feature extraction. There are two basic approaches to the computation of principal components:
(1) Batch methods – include the method of Eigen decomposition and the method of singular value decomposition (SVD)
(2) Adaptive methods – are mainly done by neural networks.
In my project, batch method is used. The main purpose of PCA is to help describe the variance–covariance structure of the data with the use of a few linear combinations of the original variables. The main concerning thing about PCA is that it utilizes only the global information of face images, this method is not very effective for different facial expressions.
(3) Linear Discriminant Analysis (LDA)
Linear discriminant analysis (LDA) is usually used in pattern and statistic recognition to help find a linear combination of the features. The combination can then be used as a linear classifier or used for dimensionality reduction purposes. LDA attempts to try model the difference between the classes of data.
However, PCA on the other hand, does not consider any differences in class, and the factor analysis builds the feature combinations based on differences rather than similarities.