An Intelligent Face Recognition System Using Eigen feature Approach for Crime Investigation

This paper presents a new approach for facial eigenfeature regularization and extraction. Image space spanned by the eigenvectors of the within-class scatter matrix is decomposed into three subspaces. Eigenfeatures are regularized differently in these subspaces based on an eigenspectrum model. This alleviates the problem of unreliable small and zero eigenvalues caused by noise and the limited number of training samples. It also enables discriminant evaluation to be performed in the full dimension of the image data. Feature extraction or dimensionality reduction occurs only at the final stage after the discriminant assessment. Experiments comparing the proposed approach with some other popular subspace methods on the FERET, ORL databases show that our method consistently outperforms others.