Pattern Recognition, 4th Edition

6.3. The Karhunen–Loève Transform

In Section 5.8 the problem of the linear transformation of a feature vector was considered in the spirit of linear discriminant analysis (LDA). The class labels of the feature vectors were assumed known, and this information was optimally exploited to compute the transformation matrix. The linear transform task will also be considered in this section but from a different perspective. Here, the computation of the transformation matrix will exploit the statistical information describing the data, and it will take place in an unsupervised mode. The Karhunen–Loève transform or principal component analysis (PCA), as it is also known, is one of the most popular methods for feature generation and dimensionality reduction ...