A preview of this full-text is provided by Optica Publishing Group.
Content available from Applied Optics
This content is subject to copyright. Terms and conditions apply.
Algorithm for computation of Zernike polynomials
expansion coefficients
Aluizio Prata, Jr., and W. V. T. Rusch
A numerically efficient algorithm for expanding a function in a series of Zernike polynomials is presented.
The algorithm evaluates the expansion coefficients through the standard 2-D integration formula derived
from the Zernike polynomials' orthogonal properties. Quadratic approximations are used along with the
function to be expanded to eliminate the computational problems associated with integrating the oscillatory
behavior of the Zernike polynomials. This yields a procedure that is both fast and numerically accurate.
Comparisons are made between the proposed scheme and a procedure using a nested 2-D Simpson's
integration rule. The results show that typically at least a fourfold improvement in computational speed can
be expected in practical use.
1. Introduction
In the analysis and evaluation of optical systems
with circular pupils it is sometimes convenient to ex-
pand a suitable optical wavefront function in a series of
Zernike polynomials.14 Two schemes for doing this
numerically have been discussed in the literature,
namely, the matrix inversion and the 2-D integration
methods.5 6 Both schemes employ discrete values of
the wavefront function taken at close enough spaced
points over the pupil area so that all the function
relevant features are sampled. In the matrix inversion
method the Zernike expansion is forced to be equal to
the wavefront function at the sampled values; this
generates a linear system of equations that is then
solved for the expansion coefficients. In the integra-
tion method each expansion coefficient is computed by
integrating the wavefront function times the corre-
sponding Zernike polynomial over the pupil area.
Both methods have their own virtues and shortcom-
ings. In particular, they are usually computationally
inefficient since the first has to solve a linear system of
equations and the second has to integrate over the
highly oscillatory Zernike polynomials. This work
deals only with the integration method, for which a
numerical algorithm that is both computationally effi-
cient and accurate is developed.
The authors are with University of Southern California,
Depart-
ment of Electrical Engineering, Los Angeles, California 90089.
Received 27 June 1988.
0003-6935/89/040749-06$02.00/0.
© 1989 Optical Society of America.
The algorithm presented below works by using sec-
ond-degree polynomial approximations in association
with the function to be expanded; the expansion coeffi-
cient integrals can then be evaluated analytically, and
the problems caused by the oscillatory behavior of the
Zernike polynomials are eliminated. The algorithm so
obtained has the important characteristic that the
number of function evaluations required to compute
the expansion coefficients are determined solely by the
number of samples required to characterize the func-
tion behavior, being independent of the order of the
expansion term being computed. This is the property
responsible for its accuracy and speed.
11. Zernike Polynomials
Since the Zernike polynomials have been extensive-
ly discussed in the literature, a complete presentation
of their characteristics is not necessary. Then, instead
of repeating results available elsewhere, in what fol-
lows we concentrate only on the few specific properties
relevant to this work.
Several slightly different Zernike polynomial nor-
malizations have been used in the literature. Without
loss in generality, here only the original normalization
(and corresponding definition) of the Zernike polyno-
mials will be adopted.14 According to it the odd and
even Zernike polynomials are given by
0UT(p,) sin
= R'7(p) mk),
e~rn(P,cp) 11 Cos (1)
respectively, with the radial functions R'(p) defined as
(n-m)/2 (n - ) ! n 21
RI7(p) = (-1)1- (nP- 2)
1![(n + m)12
-l]![(n -m)/2 -]!pf 2()
15 February 1989 / Vol. 28, No. 4/ APPLIED OPTICS 749