Adaptive Optics for Vision Science

Chapter 10.6.5 - Uses Of Deconvolution

10.6.5   Uses of Deconvolution

The results in the previous section illustrate a significant contrast enhancement
of features in the deconvolved image (in this case, of the retinal cones).
Deconvolution reduces the effect of the uncompensated components of the
AO correction in the PSFs to enhance the high spatial-frequency structure in
the image. Thus it is very good for morphological identification. It also turns
out to be helpful for quantitative measurements of the radiometry of retinal
structures in the images. Analyses by Christou, Roorda, and Williams have
demonstrated that retinal cone classification is improved with deconvolution,
as shown by the absorptance plots in Figure 10.19 [59].


10.6.6   Summary

Adaptive optics imaging significantly improves structural details in the retinal
image. However, uncorrected wavefront errors produce some residual blurring
that can be removed by image postprocessing techniques, such as deconvolution.
Linear deconvolution solves for the Fourier components of the
object irradiance, but these then have to be inverted to obtain the image that
introduces a number of problems in the presence of noise. An alternative
approach is the nonlinear estimation of the object irradiance via statistical


FIGURE 10.19 Cone classifi cation from absorptance measurements for the synthetic images (left) compared to the deconvolved images (right) in the same subject. These plots illustrate the effectiveness of deconvolution for making quantitative measurements. The solid lines are linear fi ts to the L and M cones absorptance distributions. The L and M cones are segregated more accurately using the deconvolved images, as the two distributions are better separated. (From Christou et al. [59]. Reprinted with permission from the Optical Society of America.)


principles. This addresses some of the problems associated with inversion but
is significantly slower.

In general, the PSF is not well known and an estimate has to be used for
deconvolution. Myopic deconvolution permits the PSF to be recovered as well
as the object irradiance. This is in the form of a multiframe blind deconvolution
where multiple observations of the same target are used with the assumption
that the PSF varies between them. This proves to be a powerful technique
not only for improving the spatial structure in the image but also the radiometric
(intensity) measurements from the image.

Acknowldgments   This work has been supported in part by the National
Science Foundation Science and Technology Center for Adaptive Optics,
managed by the University of California at Santa Cruz under cooperative
agreement No. AST-9876783. Financial support was also provided
by the National Eye Institute grants EY014743 to DTM and EY014375 to
AR.

UNLIMITED FREE
ACCESS
TO THE WORLD'S BEST IDEAS

SUBMIT
Already a GlobalSpec user? Log in.

This is embarrasing...

An error occurred while processing the form. Please try again in a few minutes.

Customize Your GlobalSpec Experience

Category: Interferometers
Finish!
Privacy Policy

This is embarrasing...

An error occurred while processing the form. Please try again in a few minutes.