Optica Applicata, Vol. X , No. 4, 1980
On the physical interpretation of results
in coherent imaging of diffuse objects
R . D . Bahuguna, K . K . Gu pta, K . Singh
D e p a rtm e n t of P h y sics, In d ian In s titu te of Technology, New D elhi, In d ia.
A p h y sical in te rp re ta tio n of th e m a th e m a tica l form u lation of coh eren t im aging of diffuse o b je cts under th e usual assum ption of uniform sp a tia l d istrib u tion of p oin t s ca tte re rs is given. This in te rp re ta tio n also rev eals th a t th e expression for th e au to co rre la tio n of in te n sity , as d erived b y E n lo e, is ch arged w ith a slight error.
En l o e [ 1 ] considered the coherent imaging of a diffuse surface with any
arbitrary average number N of the point scatterers per unit area and gave an extensive mathematical formulation. This anylysis was subsequently used by Ic h i o k a [2] for partially coeherent diffuse objects and has become
regarded as being very important in the studies on the statistics of lasers speckles. Physical interpretation of the resulting equations has revealed that the non-Gaussian term which apperas in the expression for the auto correlation function of the intensity distribution in the image plane is charged with an error. This error, which is due to improper combination of terms (has since then remainned there) leads to wrong physical results. A need for reinvestigation of the problem has therefore arisen.
Uniform wave of coherent
light
V Axis q Axis a» Axis
Lens Granular
Transparency
Ap ert ure Image (Far f i e l d ) Pla ne
A uniform w ave of coh eren t ligh t is in cid en t on a tra n sp a re n cy com posed of ran d o m ly d istrib u ted u n it p oin t sca tte re s. L ig h t collected b y th e a p ertu re A , p laced in th e
far-field , is im aged b y lens L on p lace P (from [1])
Schematic diagram of the optical system together with the symbols used by Enloe is shown in the figure. The complex amplitude in the image plane is given by (eq. (3 a) of [1])
where e = n(z —/)M/2, v = — (fld)xi , (o = — {f/d)yi , X is the wavelength of radiation, 0i is the relative phase of the wave scattered from the scatterer located at [x{, y{), h is the amplitude point spread function, and K is the total number of scatterers. The intensity distribution in the image plane is therefore given by K K - ■Vk/_L_J_ ** w , ^ | | 2 2 jU / + A d’ Xf + Adi)(2) k—l K K +
V yW-L+f^.^L+iLh'/JL + ^ , ^ + J(iW
Z j Zj \Xf Xd’ Xf Xd) -6i) k=£i Oncoh "i~Oioise · · · ·The first term, which is the incoherent contribution, is a fluctuating quantity and can be written as
Oncoh ( I ( N ) } + g ( v , co,xk, y k, K ) , (3) where N is the average number of scatterers per unit area of the diffuser. In the above equation <I ( N )) is the ensemble average intensity and
g(v, a), xk, yk, K) is a fluctuating term which depends on the point of observation (v, to), the position of the point scatterers (xk, yk) and the total number of scatterers K . The average value of g is zero. Because of Poisson distribution of the number of scatterers, the r.m.s. fluctuation in the number is VN which (for small N) is comparable to N. Moreover, for small N there is also a pronounced fluctuation in the incoherent contri bution from scatterers by virtue of rearrangement of their positions. On the other hand, for large N, VW is negligible compared to N and the rear rangement of the positions hardly changes the contribution. Both these effects are contained in g and can be neglected compared to <I ( N )> for large N . Hence, for large N (Gaussian case) eq. (2) is reduced to
I(v,co) = < I } + I noiae. (4)
Working on Enloe’s lines one can see that using eq. (4) the auto correlation of intensity for Gaussian case is given by
-®Gaussian(^>
CO "^-Oioise 0 > 0
W ) l 2j
i?i(0,0) r
(5)
On the 'physical interpretation... 323
which, as expected, corresponds to the first two terms of Enloe’s equation {14), where
0 0 00
Qi{v,,v) =
J j
h*(t,r)h{t-\-u,r-\-v)dtdr.— 00 — 00
In the non-Gaussian case the expression for the intensity distribution can be written as
I{v, to) = <.I } + g { v , (o,xk , y k , K ) + I noiae{v, co, xk , y k , K ) . (7) The expression for autocorrelation can subsequently be written as
non- Gaussian ( r , <) = < T >2 + R noiBe( r, t) +Rgg{r, t) + 2RgnoieG{ r, t)
/ T\2 I / r \2 \ex(r№ > t/Af) I I T) / 4\ l O 73 / -#\ /ON — C O + < 0 ---2 /n ftN--- 0 + 2 ^ n o i s e ( r ) ( 8 )
where Rnoiae has been taken from eq. (6).
E rror in Enloe’s expression for autocorrelation
Here we will see what wrong physical interpretations are inferred from Enloe’s eq. (14). Comparison of his eq. (14) with our eq. (8) gives
B „ (r , t )+ 2 B anoitt,(r, t) = 2 ^ e ,(rlXf, tlXf), (9) where
0 0 OO
q2(u,v) = j J \h(t, r)\2 \h{t+u, r-\-v)\2 dt dr.
By using standard results of short noise [3] it can further be shown that
Rgo(r, t) = <I> eArltf, tlXf)
6x( o , o ) (10)
From the above two equations we get
R(/noise( r , t )
( I ) QM(r/tf,t/Af)
2 6i( 0 , 0 ) (11a)
Further, by virtue of eqs. (8) and (10) the expression for contrast is given by
(Contrast)2 = 1 + 2 < g 2>
< i > 2 *
( lib )
The eq. (11a) shows that a speckle noise is correlated with the incoher ent fluctuations, a result which seems to be physically irrational.
Moreo-ver, the expression for contrast, as given by eq. (lib ) , also appears to be wrong, as the variance of the sum of two uncorrelated random variables is the sum of their individual variances. This indicates that Enloe’s expres sion for the autocorrelation is not correct, which was confirmed by a closer examination of his mathematical formulation. This error is due to im proper combination of terms in eq. (12) of his paper [1], where the terms for k = i — m — n appear twice instead of only once. The correct version of his eq. (12) is as follows:
K K i * = 1 m = l 1 K K
»(*
W
(K
)dK
r1dxx
J
2X
00 .7 f d x kf
dVk
J
2Y
J
2 1 J 2Y
— oo — OO — oo -oo 0 0 COy
k \ 2 U ( r + r , x m Wy
m \\ hXd
’ A / + m ! \ A f ^ M ’ A / +d j
\ fc= 1 m=l k^m ,Xk
M
Vk
' \h
*l v+r
Xk
Vk
\ r W l h s LXd
j
1 {Xd
’Xf
Xd
j
l,* l - L mHHhl \Xf + Xd’ 1/ + d + r ^ CQ+t ^ yr A/ + Xd’ A/ Xd)!]■
which leads finally to
R ( r , t ) = <I>2[ l + 1 g x f r / f f , W ) | ‘
e
\( o , o ) + ei(o,o)(1 2)
(13)
Comparing the above equation with Enloe’s eq. (14) we find that his expression contains an extra factor 2 in the last term which is the non- -Gaussian term. Using the modified equation for autocorrelation (13) the incorrect results given by eq. (11) take now the following form :
noise 0 ,
and
(Contrast)2 = 1 + ,
which now appear physically consistent.
Additional remark
I t is worth mentioning that the contrast in the equivalent far-field as determined from the modified eq. (13) is given by
(Contrast)2 = 1 + - L ,
A
where K is the average number of scatterers within the diffuser. The above equation, as expected, tallies with eq. (22) of Jakeman et al. [4].
On the physical interpretation... 325 Acknowledgements — T h is w ork w as preform ed u nd er th e “L a s e r A pp lication s P ro g ra m ” of th e In s titu te and we th a n k Professors M. S. Sodha, A . K . G h atak , and S. S. M ath ur for th e ir in te re st.
R e fe re n ce s
[1 ] EnloeL . H ., B ell S yst. T ech n . J . , 46 (1 9 6 7 ), 1 479. [2 ] Ichioka Y ., J . O pt. Soc. A m . 6 4 (1 9 7 4 ), 9 1 9 .
[3 ] Papoulis A ., Probability, Random Variables and Stochastic Processes, M cG raw -H ill, K o g a k u sh a , L t d ., 1 965, p. 357.
[4 ] Jakeman E ., McWh ir t e rJ . G ., Pu s e yP . N ., J . Opt. Soc. A m . 6 6 (1 9 7 6 ), 1 175.
Received November 7, 1979 Физическая интерпретация результатов когерентного изображения диффузионных предметов Приводится физическая интерпретация математического формализма, используемого в ко герентном изображении диффузионных объектов при обычных предположениях однород ности пространственного распределения точечных рассеивателей. Эта интерпретация объяс няет, что выражение автокорреляции интенсивности, выведенное Еп1ое, содержит незна чительные ошибки.