Abstract

The existence of a critical spatial frequency range for face recognition has been debated widely in the literature in the last decade. Ginsburg (1980) argued that the low spatial frequencies (1-4 cycles/face width) were sufficient for recognition. On the other hand, Fiorentini et al (1983) found that recognition of face images limited to spatial frequencies below 5 cycles/face width was worse than recognition of images filtered to contain only higher frequencies.

© 1991 Optical Society of America