Before classification, linear discriminant analysis is performed to reduce the number of features to a more manageable quantity. Each of the additional dimensions is a template made up of a linear combination of pixel values.
As the name implies, Linear Discriminant Analysis is a linear model for classification and dimensionality reduction. In pattern classification challenges, it's most typically employed for feature extraction.
Market researchers frequently employ discriminant analysis, a powerful statistical tool for classifying observations into two or more groups or categories. To put it another way, discriminant analysis is used to allocate things to one of several recognised groups.
When you have a categorical output variable, LDA is typically employed in classification tasks. Both binary and multi-class categorization are possible. The Gaussian Distribution of the input variables is used in a conventional LDA model.