Effects of high-pass and low-pass spatial filtering on face identification

摘要: If face images are degraded by block averaging, there is a nonlinear decline in recognition accuracy as size increases, suggesting that identification requires critical minimum range of object spatial frequencies. The faces was measured with equivalent Fourier low-pass filtering and averaging preserving the same information high-pass transformations. In Experiment 1, declined response time increased significant manner all cases spatial-frequency reduced. However, it did so at faster rate for quantized high-passed images. A second experiment controlled differences contrast found reduced but performance These data suggest preferentially supported band frequencies approximately 8-16 cycles per face; or line-based explanations were to be inadequate. discussed terms current models identification.