Sunteți pe pagina 1din 8

OPTICAL REVIEW Vol. 12, No.

3 (2005) 247–254

A Prophylactic Strategy for Global Synthesis in Lens Design


L. N. H AZRA and S. C HATTERJEE
Department of Applied Physics, University of Calcutta, 92 APC Road, Calcutta 700 009, India
(Received October 1, 2004; Accepted March 31, 2005)

Prefaced with a brief description of the state of the art in computer aided lens design optimization, this paper reports a
prophylactic strategy to obtain global or quasiglobal optimum solutions for this nondeterministic polynomial (NP) time
hard problem. For the synthesis of multicomponent lens systems, a top down approach is employed first to obtain
paraxial and central aberration targets for the constituent thin components. Central aberrations of a component are the
aberrations when stop is on the component. Suitable structural designs for the individual components are determined by
evolutionary programming. Overall systems synthesized with these quasiglobally optimum components are reoptimized
by local optimization algorithms. Illustrative numerical results on the second step of the design procedure are
presented. # 2005 The Optical Society of Japan
Key words: Optical design, lens design, global optimization, structural design, computer aided lens design, multicomponent lens systems
synthesis and evolutionary programming

1. Introduction entire lens design procedure. The second asserts that lens
design inevitably requires qualitative judgements and com-
Optical lens design involves determination of the con-
promises to be made and hence that the computer should be
structional parameters, e.g., curvatures, aspheric coefficients,
regarded as a tool capable of presenting the designer with
clear diameters, thicknesses, separations, glass materials,
provisional solutions only.’’ ‘‘The first philosophy suggests
etc. for the constituent lens elements of a lens system that
the establishment of a universal merit function and the
can achieve a prespecified set of paraxial and image quality
development of a closed procedure for minimizing its value
measures.1–7) Except for some simple cases, the problem has
without violating physical constraints. The merit function
no analytical solution. Although Hamiltonian treatment
may involve fairly extensive calculation, since only the
provides highly useful insight on the anatomy of lens
single sequence of iterations necessary for its minimization
systems, its role in practical synthesis of such a system is
is to be performed. The second philosophy suggests the
still limited. As a result, often one comes across unwanted
development of a flexible procedure which allows the
complexities and over-dependence on heuristics and empiri-
designer to exercise individual control over a number of
cism in practice. Notwithstanding the benefits accruing from
image characteristics. The characteristics should be easily
the adoption of a Lagrangian approach to tackle practical
calculable since several sequences of iterations may be
problems, efficient use of the analytical tools of Hamiltonian
performed as the designer changes his estimates of the
treatment, particularly at the conceptual structural design
relative importance of various characteristics.’’ Although
phase of a lens design problem, can lead to significantly
debate on the two philosophies continues, most practicing
improved designs.8) Indeed, global synthesis of lens designs
lens designers hold the second view.
calls for a strategy that makes a judicious combination of
anatomical strengths of Hamiltonian methods with the
3. Local and Global Optimization
diagnostic and therapeutic powers of the Lagrangian
approach. The aim of this paper is to present a systematic The merit function referred to above represents departure
method for global synthesis of lens design. The method is of the achieved quality from the desired quality of a system
based upon a strategy as envisaged above, and makes and should necessarily involve a set of system performance
appropriate use of analytical techniques and optimization functions which need to be large enough to define
methodologies in different stages of implementation. reasonably well system quality over the whole field, small
enough to avoid redundancy and simple enough to be
2. Philosophies Underlying the Formulation of Lens calculated rapidly. Obviously, selection of ray fans and
Design Optimization choice of weightings for members of the set defines the merit
function which is optimized by computer software packages.
The process of designing a lens system requires an
The default options for merit functions available with
enormous amount of numerical calculation. Use of digital
software packages for lens design are not always appropri-
computers with phenomenal number crunching capability
ate; it is common knowledge that reformulation of the merit
has greatly facilitated the process. Nevertheless, two distinct
function often leads to dramatic improvements in system
philosophies are evident on the approach to use computers in
performance.
lens design. As early as 1963, Spencer reported:9) ‘‘The first
Mathematically, the problem of optical design can be
of them asserts that the lens design problem can be
formulated as a constrained nonlinear optimization problem
completely stated in explicit mathematical terms and hence
in a multivariate hyperspace of the construction variables.
that a computer can reasonably be expected to carry out the
Usually rigid constraints are handled by the technique of

Corresponding author. E-mail address: lnhazra@sify.com Lagrange multipliers, whereas flexible constraints are dealt

247
248 OPTICAL REVIEW Vol. 12, No. 3 (2005) L. N. HAZRA and S. C HATTERJEE

NONLINEAR OPTIMIZATION METHODS

Function Evaluation only Calculus Based


No Derivatives

No Single Merit Function


Univariate Conjugate
Step by Step Directions

Simultaneous
Simplex Nonlinear Inequalities

Adaptive
Unique Merit Function Optimization

Descent Methods
Least Squares

‘Small ‘Large Steepest Variable Metric


Residual Residual’ Descent

Conjugate DFP
Undamped Gradient Method
Damped Least Squares

Nonmonotone Additive
Line Search Damping
Orthonormal
Monotone Multiplicative Preferred Optimization
Line Search Damping Number
Damping
Homogeneous
Second Derivative

Fig. 1. Hierarchy of nonlinear optimization methods used for lens design.

with with the help of penalty or barrier functions. An methods based on least squares are the most versatile from
arbitrary selection of constraints is likely to force the the point of view of speed of convergence and for avoiding
configuration hyperspace further away from the desirable undue stagnation problems.13,14) All commercially available
convex shape so that the effective search for the optimum optical design software uses this method as the bulwark of
excludes many useful optima in practice. Handling of the its optimization operation.
constraints is a tricky issue, and no single approach can be In general, multimodality of the objective function (more
expected to yield the desired results in all cases. commonly known as merit function in optics literature) of
Figure 1 presents the hierarchy of the available nonlinear the lens design problem poses a serious challenge in
optimization procedures applicable for locating the local selection of suitable starting designs that can yield globally
optimum in the neighbourhood of a given starting point in a or quasiglobally optimum solutions for the problem at hand.
constrained multivariate hyperspace of system variables. Whereas for conventional optical systems enough ‘prior art’
Over the years almost all of them have been experimented has emerged out of in-depth investigations in the field
with for tackling lens design optimization problems.10–12) carried out for a long period of time, and this has somewhat
Out of the large number of available methods, no single one lessened the problem in practice, no such guide exists for
can be chosen to be the most appropriate for all problems. systems involving unconventional optical elements and
Nevertheless, it has been observed that the optimization devices, viz., diffractive/binary optical elements, gradient
OPTICAL REVIEW Vol. 12, No. 3 (2005) L. N. HAZRA and S. C HATTERJEE 249

Table 1. Time to perform a grid search in N dimensions. freedom into a set of subproblems with considerably lower
N Time degrees of freedom. The number of variables involved in
1 10 ms
tackling the reduction procedure, and in working out the
5 100 ms
structure of the individual subproblems is such that
10 2.78 h
stochastic global optimization procedures can be adopted
15 31.7 years
for solving them. As strict adherence to aberration targets at
20 31.7 million years
the thin lens design stage is not very meaningful, a certain
25 0.317 billion (1012 ) years
number of these targets are allowed to ‘‘float’’, and often a
set of quasiglobal optima need to be taken into consid-
eration. The set of globally or quasiglobally optimal solu-
tions obtained are combined suitably to obtain a set of useful
index elements, asymmetrical optical systems, integrated solutions for the design problem. The overall systems are
optics, conformal optics, etc. separately reoptimized with the help of fast local optimiza-
A few interesting investigations have recently been tion algorithms operating with finite aberration based merit
reported to tackle this problem. Some of these are essentially function and the final solution can be picked from among
hill climbing strategies, e.g., the use of dynamic merit them on the basis of other desirable properties.
function,15) partitioning of the configuration space16) or the The basic flow diagram pertaining to this strategy is given
use of escape function;17,18) they tend to modify the local in Fig. 2, which displays the pertinent features. The
optimization algorithms so that the search is extended procedure involves seven stages. Starting from the problem
beyond the immediate neighbourhood of a local optimum. requirements (stage I), one or more prospective thin lens
Although these techniques can provide immediate remedies layouts are worked out by analytical techniques to satisfy the
to undue stagnation problems in some cases, it is obvious Gaussian specifications and the requirements for Petzval
that the chances of obtaining globally or quasiglobally curvature (stage II). Stage III involves determining the
optimum solutions are remote. On the other hand, few optimal central aberration requirements for individual
reports have appeared on the direct application of global components in a layout to satisfy aberrational requirements
optimization algorithms.19) But the inherent curse of dimen- for the overall thin layout. It is followed by stage IV where
sionality of this NP-hard problem can easily rule out the the optimum structures for individual components are
possibility of success of any deterministic global optimiza- worked out so that both Gaussian and aberrational require-
tion algorithm in this venture, except for simple cases. Even ments are fulfilled for that particular component. At this
moderately complex lens design problems have about 20–60 stage, the search for the optimum structure is carried out
variables along with a nearly equal number of constraints. from a set of preferred optical glasses. Note that stage IV is
Assuming 10 samples in each dimension are adequate to to be worked on separately for each component of the thin
locate the global optimum, and each merit function evalua- lens layouts. Both stages III and IV make use of global
tion requires 1 ms (both of which are underestimates for any optimization procedures in practical implementation. The
worthwhile lens design problem) the time required to thin components of the candidate thin lens layouts are
perform a grid search for a 25-dimensional problem is appropriately thickened (stage V) by using analytical tech-
0.317 billion (1012 ) years! (Table 1)20) niques that ensure minimal changes in Gaussian and aberra-
Therefore, attempts are currently being made with tional characteristics of the overall system. Each member of
stochastic global optimization techniques to tackle the the set of possible solutions for the required system obtained
problem. The procedures being experimented with are: thereby is reoptimized in its immediate neighbourhood by
simulated annealing, genetic algorithms, neural nets and available local optimization techniques, e.g., the damped
different combinations of the same. Preliminary results with least squares, etc. (stage VI). In stage VII, the globally
a few variable systems are quite encouraging, but their synthesized solution for the given problem is identified from
practical usefulness in cases with even moderately large the set of available options from stage VI. It is based on
degrees of freedom is yet to be established. This observation considerations like cost, weight, ease of manufacture,
stems from reported demonstrations with somewhat large ergonomics, optomechanical compatibility, etc.
degrees of freedom systems, where a part of the degrees of
freedom has had to be frozen to keep the operation within 5. Illustrative Results
manageable limits.21–26)
Our investigations on different aspects of this prophylactic
strategy for global synthesis were reported earlier and we
4. A Prophylactic Approach Towards Global Synthesis
refer to them for details.27–36) In what follows we present
In order to reduce the role of heuristics and wild some representative results on the determination of optimum
empiricism in the design of conventional lens systems, and structure for individual components from a set of preferred
to develop a systematic procedure for tackling the design optical glasses with the help of our improved version of a
problem of unconventional optics, an alternative strategy is stochastic optimization algorithm based on evolutionary
being worked out. This strategy makes an effective use of programming. Details of this algorithm have been presented
the thin lens aberration theory and paraxial analysis in in two recent publications.37,38)
reducing a formidable design problem with large degrees of At the subproblem level, an individual thin component is
250 OPTICAL REVIEW Vol. 12, No. 3 (2005) L. N. HAZRA and S. C HATTERJEE

Object 1 stop j N Image


(a)

Stage Implementation
procedure
I Problem
Requirements

Thin lens layout to satisfy Gaussian Analytical


II specifications & Petzval curvature
techniques

Determination of optimum central


aberration requirements for the
individual components in order to Global
III
satisfy aberrational requirements Optimization
for the overall system

Determination of optimum structure


IV for individual components from a set Global
of preferred optical glasses Optimization

Thickening of the thin components


with minimal changes in Gaussian
V and aberrational characteristics of Analytical
the overall system Techniques

Reoptimization of quasiglobal
optimum starting solutions in their
Local
VI Optimization
immediate neighbourhood

Identification of the desired globally


VII synthesized solution by the designer

(b)

Fig. 2. (a) An N-component lens system. (b) Basic flowchart for prophylactic lens design.

characterized by its power K, the passage of the paraxial aberrations, respectively. In our treatment the Seidel
marginal ray originating from the center of the object and coefficients are suitably scaled to obtain the primary wave
the paraxial invariant H of the system. Over and above this aberration coefficients.
gaussian requirement, the component has three central Figure 3 gives a brief schematic layout for the thin
aberration targets, namely the primary spherical aberration component and the mode of search for practical synthesis of
ðSI =8ÞT , the primary coma ðSIIc =2ÞT and the longitudinal the component. The latter commences from a singlet and is
chromatic aberration ðCL =2ÞT . Central aberrations refer to subsequently extended to structures with increasing com-
the corresponding aberrations with stop on the component. plexity, e.g., a cemented doublet, a broken contact doublet,
SI , SIIc and CL are the Seidel coefficients for the spherical combination of a cemented doublet and a singlet, a cemented
aberration, the central coma and the longitudinal chromatic triplet etc.
OPTICAL REVIEW Vol. 12, No. 3 (2005) L. N. HAZRA and S. C HATTERJEE 251

1.80

1.75

Refractive index
1.70
h
1.65

1.60

1.55

1.50
(a)
1.45
70 60 50 40 30 20

V number

Fig. 4. n–V diagram of 64 preferred optical glasses from Schott


Glass Catalog.
……

In our search, an admissible type of optical glasses for any


element of the component is treated as a discrete variable
with 64 options in a list of 64 optical glasses, selected from
Singlet Cemented Broken Contact Cemented Cemented the list of preferred optical glasses given in the Schott Glass
Doublet
Doublet Doublet
+
Triplet
Catalog.39) Figure 4 gives the corresponding (n–V) diagram.
Singlet It may be noted that the V–number of a glass is related to its
(b) dispersion n by V ¼ ½ðn  1Þ=n. For all glasses, the
operating wavelength is the Helium d line (587.6 nm), and
Fig. 3. (a) An individual thin component with paraxial targets: the lower and upper wavelengths for dispersion are the
h, K, H, and central aberration targets ðSI =8ÞT , ðSIIc =2ÞT and Hydrogen F (486.1 nm) and C (656.3 nm) lines, respectively.
ðCL =2ÞT . (b) Synthesis of a component by structures of increasing Depending upon the envisaged structure for a thin
complexity.

Table 2. A selected list of structural designs for an individual component with object at infinity; h ¼ 50 mm; K ¼
0:0025 mm1 ; H ¼ 0:218 mm. Central aberration targets: ðSI =8ÞT ¼ 2, ðSIIc =2ÞT ¼ 5, ðCL =2ÞT ¼ 0.  ¼ 0:5876 mm.
Chromatic range: 0.4861–0.6563 mm.
Singlet: Not possible
Cemented Doublet
Curvature Glass Aberrations in 
C1 C2 ¼ C3 C4 Gl 1 Gl 2 SI =8 SIIC =2 CL =2
1 0.0061 0:0116 0.0017 BaF52 F1 2.43 4.78 0:03
2 0.0089 0.0164 0.0027 F8 ZKN7 1.7 5.07 0:19

Broken Contact Doublet


Curvature Glass Aberrations in 
C1 C2 C3 C4 Gl 1 Gl 2 SI =8 SIIC =2 CL =2
1 0.0059 0:0022 0:0028 0.0004 SK5 SF1 2.04 4.93 0:12
2 0.0060 0:0028 0:0032 0.0001 BaK4 SF3 1.83 4.60 0.07

Cemented Doublet + Singlet


Curvature Glass Aberrations in 
C1 C2 ¼ C3 C4 C5 C6 Gl 1 Gl 2 Gl 3 SI =8 SIIC =2 CL =2
1 0:0069 0.0019 0:0047 0.0042 0:0008 F5 LaKN13 SK16 2.02 5.08 0.04
2 0.0075 0.0133 0.0019 0:001 0:0001 BaSF 51 SK16 SK2 2.03 4.87 0.11

Cemented Triplet
Curvature Glass Aberrations in 
C1 C2 ¼ C3 C4 ¼ C5 C6 Gl 1 Gl 2 Gl 3 SI =8 SIIC =2 CL =2
1 0.0060 0:0060 0.0020 0.0007 BaLF4 BaSF2 BaF51 1.84 4.78 0:05
2 0.0082 0.0149 0.0083 0.0036 F4 BK1 LaKN9 2.04 5.09 0.28
252 OPTICAL REVIEW Vol. 12, No. 3 (2005) L. N. HAZRA and S. C HATTERJEE

Table 3. A selected list of structural designs for an individual component with object at infinity; h ¼ 50 mm; K ¼
0:0025 mm1 ; H ¼ 0:218 mm. Central aberration targets: ðSI =8ÞT ¼ 5, ðSIIc =2ÞT ¼ 4, ðCL =2ÞT ¼ 2.  ¼ 0:5876
mm. Chromatic range: 0.4861–0.6563 mm.
Singlet: Not possible
Cemented Doublet
Curvature Glass Aberrations in 
C1 C2 ¼ C3 C4 Gl 1 Gl 2 SI =8 SIIC =2 CL =2
1 0.0028 0.0094 0:0018 BaSF51 LaKN7 5.10 4:29 2:15
2 0.0030 0.0116 0:0020 F8 BaK5 5.10 4:25 2:00

Broken Contact Doublet


Curvature Glass Aberrations in 
C1 C2 C3 C4 Gl 1 Gl 2 SI =8 SIIC =2 CL =2
1 0.0033 0.0078 0.0077 0:0016 SF19 SK5 4.87 3:67 2:20
2 0.0032 0.0072 0.0070 0:0017 SF1 SK8 4.84 4:11 1:82

Cemented Doublet + Singlet


Curvature Glass Aberrations in 
C1 C2 ¼ C3 C4 C5 C6 Gl 1 Gl 2 Gl 3 SI =8 SIIC =2 CL =2
1 0:0033 0:0152 0:0060 0.0041 0.0020 SSKN8 BaSF2 BaK1 5.17 4:15 2:14
2 0:0033 0:0184 0:0036 0.0017 0:0032 SK3 BaF8 BK7 5.12 4:20 2:08

Cemented Triplet
Curvature Glass Aberrations in 
C1 C2 ¼ C3 C4 ¼ C5 C6 Gl 1 Gl 2 Gl 3 SI =8 SIIC =2 CL =2
1 0.0033 0.0118 0.0014 0:0020 F8 BaK5 K7 4.98 3:81 2:16
2 0.0017 0.0005 0:0108 0:0032 SK16 BaK4 F4 4.95 3:91 2:12

component, the latter consists of one or more thin lens in each of them approximately equal to 0.01.
elements. Therefore, in general, no shape variable for the Tables 2–4 present three selected lists of structural
thin component can be defined, and shapes of the constituent designs for three different sets of central aberration targets
thin elements are to be taken into consideration. Each for an individual component with the same paraxial
element of the component is characterized by its shape specifications: h ¼ 50 mm; K ¼ 0:0025 mm1 ; object at
variable and normalized power. The shape variable X j and infinity and H ¼ 0:218 mm., where H is the Lagrange
the normalized power kj of the jth element are given by invariant of the system, K is the power of the component and
c1 þ c2 h is the height of the paraxial marginal ray on the
Xj ¼ and kj ¼ k j =K component. The convenient amenability of our stochastic
c1  c2
evolutionary algorithm in providing ‘floating’ of aberrations
where c1 and c2 are curvatures of the first and second is evident from the values of achieved aberration targets.
surfaces of the lens element and k j is the power of the jth Allowing a higher degree of floating, many more candidate
element. Note that K is power of the component which solutions may be obtained in each type. Table 5 gives a list
consists of J thin elements in contact. of nine solutions for achromatic aplanatic cemented doublets
X
J for the same paraxial targets. No relaxation in paraxial
K¼ kj targets is, however, permitted in any of the solutions.
j¼1
6. Concluding Remarks
where J ¼ 1; 2; 3; . . . refer to thin singlet, doublet, triplet,
. . .respectively. In the stage IV [Fig. 2(b)], the search Acceptability of a solution as a quasiglobal solution for
domain for X j and kj for all elements of a component is the problem at hand is to be based on practical viability that,
confined within the limits in turn, needs to be ascertained on considerations like
manufacturing constraints, possible occurrence of higher
jX j j  5:0 and jkj j  5:0 for j ¼ 1; . . . ; J
order aberrations etc.
to reduce the possibility of infeasible and unacceptable It is important to note that this prophylactic strategy for
curvatures in the lens elements. Binary coding for both these synthesis of lens systems obviates the need for heuristic
variables is implemented in a manner that sets the precision preselection of either the structure for the lens system or the
OPTICAL REVIEW Vol. 12, No. 3 (2005) L. N. H AZRA and S. C HATTERJEE 253

Table 4. A selected list of structural designs for an individual component with object at infinity; h ¼ 50 mm; K ¼
0:0025 mm1 ; H ¼ 0:218 mm. Central aberration targets: ðSI =8ÞT ¼ 4, ðSIIc =2ÞT ¼ 3, ðCL =2ÞT ¼ 0.  ¼ 0:5876
mm. Chromatic range: 0.4861–0.6563 mm.
Singlet: Not possible
Cemented Doublet
Curvature Glass Aberrations in 
C1 C2 ¼ C3 C4 Gl 1 Gl 2 SI =8 SIIC =2 CL =2
1 0.0077 0.0135 0.0015 F1 K7 4:08 3.03 0.20
2 0.0053 0:0058 0.0001 BaLF4 BaF13 3:76 3.14 0:08

Broken Contact Doublet


Curvature Glass Aberrations in 
C1 C2 C3 C4 Gl 1 Gl 2 SI =8 SIIC =2 CL =2
1 0.0049 0:0023 0:0027 0.0006 LaK24 SF3 4:23 2.86 0:09
2 0.0050 0:0050 0:0054 0.0013 LaKN7 F2 4:27 2.80 0:16

Cemented Doublet + Singlet


Curvature Glass Aberrations in 
C1 C2 ¼ C3 C4 C5 C6 Gl 1 Gl 2 Gl 3 SI =8 SIIC =2 CL =2
1 0.0072 0.0133 0.0006 0:0003 0.0001 F8 BK1 BaSF51 3:9 2.90 0.00
2 0:0096 0:0009 0:0097 0.0045 0.0002 SF2 BaF3 LaKN13 4:1 3.13 0.23

Cemented Triplet
Curvature Glass Aberrations in 
C1 C2 ¼ C3 C4 ¼ C5 C6 Gl 1 Gl 2 Gl 3 SI =8 SIIC =2 CL =2
1 0.0055 0:0048 0:0030 0:0007 K7 SF3 F2 4:08 3.02 0.07
2 0.0054 0:0060 0.0083 0:0004 BK7 F4 F2 3:90 2.94 0:25

Table 5. A selected list of structural designs for an aplanatic, achromatic cemented doublet with object at infinity;
h ¼ 50 mm; K ¼ 0:0025 mm1 ; H ¼ 0:218 mm. Central aberration targets: ðSI =8ÞT ¼ ðSIIc =2ÞT ¼ ðCL =2ÞT ¼ 0.  ¼
0:5876. Chromatic range: 0.4861–0.6563 mm.
Achieved aberrations
Solution Curvatures Glass
(in wavelengths)
numbers
C1 C2 ¼ C3 C4 Glass1 Glass2 SI =8 SIIC =2 CL =2
1 0.0048 0.0137 0.000 F8 BaK1 0.00 0.00 0.00
2 0.0051 0.0095 0.0002 SF3 SSKN8 0.01 0:04 0.02
3 0.0054 0.0096 0.0001 SF1 BaLF4 0:02 0.06 0.02
4 0.0038 0:0066 0:0008 SK3 BaSF2 0.06 0:05 0:06
5 0.0059 0.0099 0:0002 SF19 BK1 0.05 0:12 0:07
6 0.0051 0.0106 0.0000 SF2 BaK1 0.03 0:08 0:04
7 0.0036 0:0072 0:0004 BaF50 SF1 0.15 0:11 0:09
8 0.0064 0.0098 0:0002 SF3 K10 0:06 0.18 0.08
9 0.0053 0.0103 0.0001 SF19 BaK1 0:10 0.11 0.07

optical materials for the constituent lens elements. References


Finally, it is instructive to look into the inherent similarity 1) R. R. Shannon: The Art and Science of Optical Design (Cambridge
between the practice of medicine and that of optical design University Press, Cambridge, 1997).
(Table 6). In the former, the role of prophylaxis is well 2) M. J. Kidger: Fundamental Optical Design (SPIE Press, Bellingham,
established. Success in ventures of complex optical design Washington, D.C., 2002).
3) M. Laikin: Lens Design (Marcel Dekker, NewYork, 1991).
catering to the exigencies of modern science and technology 4) W. J. Smith: Modern Optical Engineering (McGraw Hill, New York,
calls for suitable prophylactic procedures in the optical 2000).
design. 5) D. P. Feder: Appl. Opt. 2 (1963) 1209.
254 OPTICAL REVIEW Vol. 12, No. 3 (2005) L. N. HAZRA and S. C HATTERJEE

Table 6. Similarity between practice of medicine and optical 12) M. J. Hayford: Proc. SPIE 531 (1985) 68.
design. 13) M. J. Kidger: Opt. Eng. 32 (1993) 1731.
14) J. Basu and L. N. Hazra: Opt. Eng. 33 (1994) 4060.
Medicine Optical design 15) S. Banerjee and L. N. Hazra: Opt. Eng. 36 (1997) 3111.
1. Anatomy a. Ray tracing 16) S. Banerjee and L. N. Hazra: Proc. SPIE 3482 (1998) 126.
b. Basic laws of image formation 17) M. Isshiki, H. Ono, K. Hiraga, J. Ishikawa and S. Nakadate: Opt. Rev.
2 (1995) 463.
18) M. Isshiki: Proc. SPIE 3482 (1998) 104.
2. Diagnosis a. Gaussian optics 19) D. Sturlesi and D. C. O’shea: Opt. Eng. 30 (1991) 207.
b. Seidel optics 20) L. N. Hazra: Perspectives in Modern Optics & Optical Instrumenta-
c. Finite aberrations tion (Anita Publication, New Delhi, 2002) p. 30.
d. Spot diagram 21) V. K. Viswanathan, I. O. Bohachevsky and T. P. Cotter: Proc. SPIE
554 (1985) 10.
e. P.S.F.; O.T.F.; E.S.F. . .
22) A. E. W. Jones and G. W. Forbes: in Proc. OSA International Optical
Design Conf., ed. G. W. Forbes (Optical Society of America,
3. Therapy a. Knowledge based adhoc tools, e.g., stop shift, Washington, D.C., 1994) Vol. 22, p. 42.
change of power between two components or 23) D. C. Van Leijenhorst, C. B. Lucasius and J. M. Thijssen: BioSystems
surfaces, etc. 37 (1996) 177.
24) X. Chen and K. Yamamoto: J. Mod. Opt. 44 (1997) 1693.
b. Software packages for optimization
25) I. Ono and S. Kobeyashi: Proc. SPIE 3482 (1998) 110.
26) S. Banerjee and L. N. Hazra: Opt. Eng. 37 (1998) 3260.
4. Prophylaxis Use of quasiglobal optimum structural design 27) L. N. Hazra: Appl. Opt. 23 (1984) 4440.
28) L. N. Hazra and A. K. Samui: Appl. Opt. 25 (1986) 3721.
29) S. Banerjee and L. N. Hazra: Proc. SPIE 3430 (1998) 175.
30) L. N. Hazra and C. A. Delisle: J. Opt. Soc. Am. A 15 (1998) 945.
31) S. Banerjee and L. N. Hazra: Proc. SPIE 3737 (1999) 172.
6) D. Malacara and Z. Malacara: Handbook of Optical Design (Marcel 32) S. Banerjee and L. N. Hazra: in Optics and Optoelectronics: Theory,
Dekker, NewYork, 2004). Devices and Applications, eds. O. P. Nijhawan et al. (Narosa
7) R. E. Fischer and B. Tadic-Galeb: Optical System Design (McGraw Publishing, London, 1999) Vol. 1, p. 499.
Hill, NewYork, 2000). 33) S. Banerjee and L. N. Hazra: Appl. Opt. 40 (2001) 6265.
8) H. A. Buchdohl: An Introduction to Hamiltonian Optics (Cambridge 34) S. Banerjee and L. N. Hazra: J. Mod. Opt. 49 (2002) 1111.
University Press, London, 1970). 35) S. Chatterjee and L. N. Hazra: Proc. CIIC (Calcutta University Press,
9) G. H. Spencer: A Computer Oriented Automatic Lens Correction Kolkata, 2001) p. 659.
Procedure (Institute of Optics, Rochester, 1963). 36) L. N. Hazra: SERC Research Highlights (Dept. of Science &
10) T. H. Jamieson: Optimization Techniques in Lens Design (Adam Technology, Govt. of India, 2003) p. 171.
Hilger, London, 1971). 37) S. Chatterjee and L. N. Hazra: Opt. Eng. 43 (2004) 432.
11) A. K. Rigler and R. J. Pegis: in Topics in Applied Physics, ed. B. R. 38) S. Chatterjee and L. N. Hazra: J. Opt. 33 (2004) 109.
Frieden (Springer-Verlag, Berlin, 1980) Vol. 41, p. 211. 39) Optical Glass Catalog (Schott Glass, Mainz, 1999).

S-ar putea să vă placă și