Additionally, to overcome the problem of multicollinearity in
linear regression, the ridge regression and principal component
regression are widely used. Moreover, an improved principal
component regression classification (IPCRC), which removes
the mean of each image before applying principal component
analysis and drops the first principal components, can help
to eliminate lighting effects for face classification [10]. Simulation
results reveal that the IPCRC outperforms the conventional
principal component regression classification (PCRC),
the LRC, the ridge regression classification (RRC) and the discriminant-
based approaches for face recognition under illumination
changes.