Badcock D R
Department of Psychology, University of Melbourne, Parkville, Victoria, Australia.
J Exp Psychol Hum Percept Perform. 1990 Feb;16(1):217-20. doi: 10.1037//0096-1523.16.1.217.
Fourier analysis may be used to obtain the amplitude and phase spectra of images. Spectra from two images can then be recombined in an attempt to produce images with identical phase spectra but different amplitude spectra or with identical amplitude but different phase spectra. Kleiner and Banks (1987) and Kleiner (1987) used this technique to determine if the amplitude (energy) or phase characteristics of images form the basis for infant face preferences. This article describes how some of the procedures used by these researchers, for example, luminance normalization and photographic production of transparencies of images, probably prevented the images from having the desired characteristics, thus making interpretation of their studies very difficult. Additional potentially confounding issues to be addressed when repeating these or similar experiments are also discussed. These difficulties arise from the distinction between the numerical phase values obtained through the mathematical process and those that are visually effective.
傅里叶分析可用于获取图像的振幅谱和相位谱。然后，可以将来自两幅图像的谱进行重新组合，以尝试生成具有相同相位谱但不同振幅谱的图像，或者具有相同振幅但不同相位谱的图像。克莱纳和班克斯（1987年）以及克莱纳（1987年）使用这项技术来确定图像的振幅（能量）或相位特征是否构成婴儿面部偏好的基础。本文描述了这些研究人员所使用的一些程序，例如亮度归一化和图像透明片的照相制作，可能如何阻止图像具有所需的特征，从而使其研究的解释变得非常困难。文中还讨论了在重复这些或类似实验时需要解决的其他潜在混淆问题。这些困难源于通过数学过程获得的数值相位值与视觉上有效的相位值之间的差异。
