Title :
Building a representative training set based on eigenimages
Author :
Herbst, B.M. ; Muller, N.L.
Author_Institution :
Dept. of Appl. Math., Stellenbosch Univ., South Africa
Abstract :
Adding new images to the training set necessitates a recalculation of the eigenfaces as well as the representation of all the images in the database in terms of the new eigenfaces. We describe an efficient procedure for doing this and give some examples of the performance of the procedure in practice
Keywords :
covariance matrices; eigenvalues and eigenfunctions; face recognition; image reconstruction; eigenfaces; eigenimages; representative training set; Africa; Character recognition; Covariance matrix; Eigenvalues and eigenfunctions; Face recognition; Image resolution; Mathematics; Optical character recognition software; Performance evaluation; System testing;
Conference_Titel :
Pattern Recognition, 1998. Proceedings. Fourteenth International Conference on
Conference_Location :
Brisbane, Qld.
Print_ISBN :
0-8186-8512-3
DOI :
10.1109/ICPR.1998.712090