This paper addresses the problem of face recognition using independent component analysis. As the independent components (IC) are not orthogonal, to represent a face image using the determined ICs, the ICs have to be orthogonalized, where two methods, namely Gram-Schmit Method and Householder Transformation, are proposed. In addition, to find a better set of ICs for face recognition, an efficient IC selection algorithm is developed. Face images with different facial expressions, pose variations and small occlusions are selected to test the ICA face representation and the results are encouraging.
CITATION STYLE
Yuen, P. C., & Lai, J. H. (2000). Independent component analysis of face images. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 1811, pp. 545–553). Springer Verlag. https://doi.org/10.1007/3-540-45482-9_55
Mendeley helps you to discover research relevant for your work.