Factor analysis of essential facial features
Vladimir G. Ivancevic, A.K. Kaine, Brett McLindin, Jadranka Šunde · IEEE Transactions on Image Processing · 2004
We present the results of an exploratory factor analysis to determine the minimum number of facial features required for recognition. In the introduction, we provide an overview of research we have done in the area of the evaluation of facial recognition systems. In the next section, we provide a description of the facial recognition process (including signal processing operations such as face capture, normalisation, feature extraction, database comparison, decision and action). The following section provides description of the variables. We follow with our main result, where we have used a sample of 80 face images from a test database. We have preselected 20 normalized facial features and extracted six essential orthogonal factors, representing 76.5 % of the total facial feature variability. They are (in order of importance): upper-lip, eyebrow-position, nose-width, eye-position, eye/eyebrow-length and face-width.