Kernel Eigenfaces Framework for Feature Extraction and Face Recognition
Megha Kamble, Sanjay Laxmikant Nalbalwar, Swarali P. Sheth, Babasaheb Ambedkar · 2015
In this paper, three methods, namely, Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Kernel Principal Component Analysis (KPCA) are implemented successfully for feature extraction and recognition of 2-dimensional face images. PCA linearly transforms the original image space to an orthogonal eigenspace with reduced dimensionality whereas LDA performs linear transformation by maximizing the ratio of between class variance and within class variance. Non-linear subspace derived using kernel method, by adopting Gaussian kernel, has been found to be superior compared to linear subspaces. The ORL and JAFFE face image databases are used to perform and test experimental analysis and results presented in this paper.