May 2008
Volume 49, Issue 13
ARVO Annual Meeting Abstract  |   May 2008
Pattern Recognition as a Retinal Image Quality Metric
Author Affiliations & Notes
  • C. D. Pinto
    Bausch & Lomb Inc, Rochester, New York
    Optical Design & Metrology,
  • A. C. Kingston
    Bausch & Lomb Inc, Rochester, New York
    Optical Design & Metrology,
  • P. Ludington
    Bausch & Lomb Inc, Rochester, New York
    Optical Design & Metrology,
  • M. S. Venkiteshwar
    Bausch & Lomb Inc, Rochester, New York
    Clinical Research Administration,
  • Footnotes
    Commercial Relationships  C.D. Pinto, Bausch & Lomb, E; A.C. Kingston, Bausch & Lomb, E; P. Ludington, Bausch & Lomb, E; M.S. Venkiteshwar, Bausch & Lomb, E.
  • Footnotes
    Support  None.
Investigative Ophthalmology & Visual Science May 2008, Vol.49, 995. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      C. D. Pinto, A. C. Kingston, P. Ludington, M. S. Venkiteshwar; Pattern Recognition as a Retinal Image Quality Metric. Invest. Ophthalmol. Vis. Sci. 2008;49(13):995. doi:

      Download citation file:

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

  • Supplements

Purpose: : To evaluate and compare a new objective image quality metric, Pattern Recognition, to the Visual Strehl (VSOTF), Neural Sharpness (NS), and Normalized MTF (SIQ) metrics and results of a psychophysical study on the perception of subjective blur.

Methods: : 25 blurred images were created by convolving an asymmetric (F) and a symmetric (+) character with the point spread function of a Liou-Brennan model eye (with and without coma) across 5 levels of defocus and spherical aberration. 15 subjects were required to rank these images from "least" to "most" blurred images.A novel, commercially available metric, Pattern Recognition, was used to provide a score from 0 to 1000 for each of the blurred characters which accounts for the phase transfer function. A score of 1000 corresponds to a diffraction limited image. Correlation analysis was performed between subjective blur ranks to Pattern Recognition ranks. The results were compared to that obtained between VSOTF, NS, and SIQ metrics to subjective blur ranks.

Results: : For the images convolved with coma on the Liou-Brennan model eye, the correlation coefficient (r) between the Pattern Recognition metric and the subjective blur rank scores from the psychophysical study was 0.874 (p<0.0001) for the symmetric ‘+’ target and 0.839 (p< 0.001) for the asymmetric ‘F’ target. For the images convolved without coma in the model eye, the correlation increased to 0.94 and 0.91 (p<0.0001) for the ‘F’ and ‘+’ targets, respectively.Correlation between commonly used metrics and subjective blur ranks will be discussed and compared to that with Pattern Recognition.

Conclusions: : For both symmetric and asymmetric characters, Pattern Recognition is significantly correlated to subjective perception of blur. Incorporating the phase of the optical transfer function by using Pattern Recognition could provide better understanding of subjective blur perception.

Keywords: imaging/image analysis: non-clinical • aberrations • visual acuity 

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.