Solving the Principal Component Analysis and Linear Discriminant Analysis problems with training Separator Dictionary algorithms

In this article, two proposed methods are presented. The first method is supervised and has a process similar to K-SVD, and the atoms will be trained in such a way that in addition to reducing the display error, resolution is also created. The second method is based on subspace clustering. By clustering based on the fact that each training data is placed under a space with limited dimension, the problem of dictionary training will be solved automatically. Finally, LDA and SVD operations as well as KLDA and KSVD dictionary training algorithms are performed on the raw and trained data and the results will be analyzed and reviewed.