December 2002
Volume 43, Issue 13
ARVO Annual Meeting Abstract  |   December 2002
What Makes Even Texture Different From Random Texture?
Author Affiliations & Notes
  • L Liu
    Vision Research Lighthouse International New York NY
  • Footnotes
    Commercial Relationships   L. Liu, None. Grant Identification: JDRF Grant #1-2001-619
Investigative Ophthalmology & Visual Science December 2002, Vol.43, 2910. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      L Liu; What Makes Even Texture Different From Random Texture? . Invest. Ophthalmol. Vis. Sci. 2002;43(13):2910.

      Download citation file:

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

  • Supplements

Abstract: : Purpose:Many studies have suggested that the pre-attentive discrimination between intact even and random textures (first pair) was based on the difference in conspicuous features, such as elongated bars. How conspicuous these features have to be in order to support pre-attentive discrimination? Methods:Even and random textures were degraded in two ways: 1) randomly deleting checks to black (deletion); 2) randomly deleting checks and then scrambling the positions of the remaining checks (scrambling). The second and third pairs of textures are examples of a 30% deletion and a 10% deletion plus a 1-check scrambling. Identically degraded even/random or random/random texture pairs were presented for 250 ms, and subjects were asked to indicate whether the pairs were made of the same or different textures. Results:Normal subjects could keep perfect discrimination when deletion was less than 40%. Greater amount of deletion caused performance to decrease. No reliable discrimination could be observed for deletion greater than 60%. When a check was allowed to move into the position of one of its deleted immediate neighbors (1-check scrambling), discrimination performance decreased even with a 5% deletion, and discrimination was impossible beyond 15% of deletion. A computer program was written to tally the lengths and positions of continuous bars in even and random textures. Intact random textures rarely had bars longer than 8 checks, but had many shorter bars. The alignment of these bars was poor. The beginning and ending positions were evenly distributed across the texture patch. Intact even textures had more longer bars and they were aligned on a small number of locations. Degradation reduced both bar length and alignment differences between even and random textures. Bar length distribution became identical at about 40% of deletion while alignment difference diminished at about 60% of deletion. Conclusion:Human subjects used both bar length and bar alignment information to discriminate even and random textures. 

Keywords: 597 texture • 509 pattern vision 

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.