Receptive field spaces and class-based generalization from a single view in face recognition

Abtraet. We describe a computational model of face recognition, which generalizes from single views of faces by taking advantage of prior experience with other faces. seen under a wider range of viewing conditions. The model represents face images by veclo~s of activities of graded overlapping receptive fields (m). It relies on high-spatial-frequency information to estimate the~viewing conditions, which are then used to normalize (via a h’ansfonnation specific for faces), and identify, the low-spatial-frequency representation of the input. The class-specific msformatian approach allows the model to replicate a series of psychophysical findings on face recognition and constitutes an advance over cmnt face-recognition methods, which are incapable of generalization from a single example.