Biometric analysis for finger vein data - Emerging Trends in Image Processing, Computer Vision, and Pattern Recognition - page 329

(2)
where S x represents the covariance matrix of the projected feature vector and tr( S x ) is the trace
of S x . To maximize the criterion in Equation (2) , the appropriate projection direction X needs
to be found. S x is introduced by
Therefore,
(3)
Deining the image covariance scater matrix, we have
(4)
We now obtain the n × n matrix of G t from all training images, where there are m training
images of size m × n . Thus, G t can be evaluated by