Principal Component and Neural Network Analyses of Face Images: Explorations into the Nature of Information Available for Classifying Faces by Sex

In this paper we present an overview of the major ndings of the principal component analysis (pca) approach to facial analysis. In a neural network or connectionist framework this approach is known as the linear autoassociator approach. Faces are represented as a weighted sum of macrofeatures (eigenvectors or eigenfaces) extracted from a cross-product matrix of face images. Using sex categorization as an illustration , we analyzed the robustness of this type of facial representation. We show that eigenvectors representing general categorical information can be estimated using a very small set of faces and that the information they convey is generalizable to new faces.

[1]  Stephen A. Ritz,et al.  Distinctive features, categorical perception, and probability learning: some applications of a neural model , 1977 .

[2]  Ashok Samal,et al.  Automatic recognition and analysis of human faces and facial expressions: a survey , 1992, Pattern Recognit..

[3]  A. Young,et al.  Cross-Domain Semantic Priming in Normal Subjects and a Prosopagnosic Patient , 1988, The Quarterly journal of experimental psychology. A, Human experimental psychology.

[4]  A. J. O'toole,et al.  Classifying faces by face and sex using an autoassociative memory trained for recognition , 1991 .

[5]  L Sirovich,et al.  Low-dimensional procedure for the characterization of human faces. , 1987, Journal of the Optical Society of America. A, Optics and image science.

[6]  Toshiyuki Sakai,et al.  Processing of multilevel pictures by computer-the case of photographs of human face , 1971 .

[7]  A. Yuille.  Deformable Templates for Face Recognition , 1991, Journal of Cognitive Neuroscience.

[8]  L. D. Harmon.  The recognition of faces. , 1973, Scientific American.

[9]  A J O'Toole,et al.  More about the Difference between Men and Women: Evidence from Linear Neural Networks and the Principal-Component Approach , 1995, Perception.

[10]  Osamu Nakamura,et al.  Identification of human faces based on isodensity maps , 1991, Pattern Recognit..

[11]  Y. Kaya,et al.  A BASIC STUDY ON HUMAN FACE RECOGNITION , 1972 .

[12]  J. Sergent.  Microgenesis of Face Perception , 1986 .

[13]  K. Cave,et al.  Can We Lose Memories of Faces? Content Specificity and Awareness in a Prosopagnosic , 1991, Journal of Cognitive Neuroscience.

[14]  Leon D. Harmon,et al.  Automatic recognition of human face profiles , 1977 .

[15]  Arthur P Ginsburg,et al.  Visual Information Processing Based on Spatial Filters Constrained by Biological Data. , 1978 .

[16]  M. Turk,et al.  Eigenfaces for Recognition , 1991, Journal of Cognitive Neuroscience.

[17]  A. Young,et al.  Childhood prosopagnosia , 1989, Brain and Cognition.

[18]  Ashok Samal,et al.  Minimum resolution for human face detection and identification , 1991, Electronic Imaging.

[19]  Hervé Abdi,et al.  A NEURAL NETWORK PRIMER , 1994 .

[20]  Alice J. O'Toole,et al.  CATEGORIZATION AND IDENTIFICATION OF HUMAN FACE IMAGES BY NEURAL NETWORKS: A REVIEW OF THE LINEAR AUTOASSOCIATIVE AND PRINCIPAL COMPONENT APPROACHES , 1994 .

[21]  Alice J. O'Toole,et al.  Connectionist models of face processing: A survey , 1994, Pattern Recognit..

[22]  A. M. Burton,et al.  What's the Difference between Men and Women? Evidence from Facial Measurement , 1993, Perception.

[23]  Teuvo Kohonen,et al.  Associative memory. A system-theoretical approach , 1977 .

[24]  H. Ellis,et al.  Parallel processing of the sex and familiarity of faces. , 1987, Canadian journal of psychology.