Introduction - If you have any usage issues, please Google them yourself
According to two kinds of training set samples separately calculated covariance matrix as produced by k-l transform matrix, and the whole of general produced by k-l matrix transformation of vector is existing translation