May 2005
Volume 46, Issue 13
ARVO Annual Meeting Abstract  |   May 2005
Rapid Visual Scene Categorization Relies Mainly on Amplitude Spectrum
Author Affiliations & Notes
  • N. Guyader
    Psychology, UCL, London, United Kingdom
  • A. Chauvin
    Faculté des arts et des sciences, Laboratoire VIC, Montréal, PQ, Canada
  • L. Bert
    LPNC, Grenoble, France
  • M. Mermillod
    LPNC, Grenoble, France
  • J. Hérault
    Ujf, LIS, Grenoble, France
  • C. Marendaz
    LPNC, Grenoble, France
  • Footnotes
    Commercial Relationships  N. Guyader, None; A. Chauvin, None; L. Bert, None; M. Mermillod, None; J. Hérault, None; C. Marendaz, None.
  • Footnotes
    Support  None.
Investigative Ophthalmology & Visual Science May 2005, Vol.46, 5642. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      N. Guyader, A. Chauvin, L. Bert, M. Mermillod, J. Hérault, C. Marendaz; Rapid Visual Scene Categorization Relies Mainly on Amplitude Spectrum . Invest. Ophthalmol. Vis. Sci. 2005;46(13):5642.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

Abstract: : Purpose: Models of the visual cortex are based on image decomposition according to the Fourier spectrum (amplitude and phase). It is commonly believed that phase information is necessary to identify a scene. But, it is also known that cortical complex cells code only the amplitude spectrum. This raises the question of knowing if these cells carry sufficient information to allow visual scene categorization. In this work, using the same experiments in computer simulation and in psychophysics, we provide arguments to show that the amplitude spectrum alone is sufficient for natural scene categorization. Methods: We used a simple model inspired by the human visual system, which whitened images and performed contrast equalization as the retinal processing. Then it decomposed images using a bank of Gabor filters. We built an experiment using a priming paradigm for which the priming image was displayed in a subliminal way. For the priming image, we used amplitude images (images with a random phase), phase images (images with a flat amplitude spectrum), chimera (images with amplitude spectrum from one category and a phase spectrum from another), original images and inverted images (the amplitude spectrum is then the same as the original image). We measured the accuracy and the reaction time of subjects/model to categorize target images as a function of the consistency between the prime and the target. The congruent condition is when primes and targets have amplitude spectrum from the same category and the incongruent condition is when prime and target have amplitude spectrum from different categories. Results: As predicted by our computational model, psychophysical data showed that the same priming effect was obtained when prime and target had similar amplitude spectrum. Conclusions: Our results fit with the observation that phase information is ignored by complex cells in low level vision and that an artificial neural network is able to categorize natural scenes on the only basis of their amplitude spectra. Visual scene categorization can be viewed as a solution to a general information–processing problem. To this end, the amplitude spectrum of a scene has a more general significance than its phase spectrum.

Keywords: scene perception • computational modeling • perception 

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.