Professional Documents
Culture Documents
Received 29 July 1996, in final form 1 April 1997, accepted for publication
15 April 1997
0957-0233/97/070764+09$19.50
c 1997 IOP Publishing Ltd
Complex modulation characterization of liquid crystal devices
pol.: polarizer
BS: beamsplitter
OUTPUT INPUT
M: mirror
LCTV: liquid
crystal
television
BS2 videoprojector
CCD
pol. M2
LCTV
λ/2
collimator
M1
He-Ne LASER
BS1 pol.
Figure 1. The experimental set-up for the phase measurement with a Mach–Zender interferometer.
scene by analysis of the optical correlation product between modulation zero), while different values are displayed on
them. the other half.
We will next prove that the LCTV characterization
results can be applied to a novel highly-efficient optical
pattern recognition design, based on a phase-only joint 3. Phase measurement method
transform correlator.
A typical fringe pattern is shown in the middle of figure 2,
where we can see two cosinusoidal interference fringes out
2. Experimental set-up
of phase, the left one (l) corresponding to the reference level
The experimental set-up used to measure the phase (gl = 0) and the right one (r) to a grey level of 64. The
modulation of the LCTV is based on a Mach–Zender pixels of the LCTV are also visible. The whole digitized
interferometer, as shown in figure 1. The collimated He–Ne picture has 512 rows and columns, as indicated.
laser beam is split by the first beamsplitter. A CCD camera Because the interference fringes are aligned in the x or
registers the interference fringes of the two plane waves horizontal direction, it is possible to represent their vertical
when the beam joins again. The phase variation produced profiles by the one-dimensional mathematical expressions
when a LCTV is placed in one arm of the interferometer that follow:
is translated onto a fringe displacement on the interference Il (y) = al (y) + bl (y) cos[2πy/P + φl (y)]
plane [9]. The λ/2 plate in the other arm compensates
Ir (y) = ar (y) + br (y) cos[2πy/P + φr (y)]
the light polarization changes as it traverses the panel, so
(y = 0 . . . N − 1). (1)
maximum contrast interference fringes are obtained.
Both the videoprojector and the camera are connected to y is a discrete variable taking N different values, where N
a PC computer through a Matrox-PIP-1024B 8-bit digitizer is the number of rows (N = 512). P is the period of the
board, where it is possible to change the input SLM image fringes, which is related to their frequency by
and to analyse the output fringe pattern. The possible grey-
level values in the computer are comprised between 0 and f = N/P . (2)
255.
The LCTV modulation depends on the orientation of the ai (y) (i = l, r) represent possible background non-
two polarizers, the position of some potentiometer controls uniformities and bi (y) are related to the local contrast of the
available on the videoprojector, labelled as colour, contrast patterns. Equations (1) work for each column and should
and brightness and, fundamentally, on the PC input image be approximately equal for all of them. In order to reduce
grey level (gl). When all these variables but the last one the non-uniformities and with the final idea of having an
remain constant, it is interesting to study the phase and average measurement, we have determined the mean value
amplitude modulation of the light traversing the device of the digitized intensity for each row, considering the
as the input grey level changes, in order to determine central 128 columns (comprised between the two vertical
the modulation of the LCTV when a grey-level image is white lines depicted in figure 2) for both the left and
displayed on it. right fringes. The corresponding distributions are shown
The method consists of the measurement of the relative in figures 2(a) and 2(b).
phase shift between several pairs of grey levels displayed on The phase shift we have to determine is given by
the panel: half of the screen always remains with a constant
grey-level value, gl = 0, which acts as a reference (phase φ = φr (y) − φl (y) = 2π 1/P (3)
765
E Martı́n-Badosa et al
Column Number
0 64 192 256 320 448 512
0 0
64 64
128 128
Row Number
Row Number
192 192
256 256
320 320
384 384
448 448
512 512
160 140 120 100 80 60 40 20 20 40 60 80 100 120 140 160
Digitized Value Digitized Value
(a) (b)
Figure 2. Digitized average interference distributions compared with fringe patterns: (a) gl = 0 (left), (b) gl = 64 (right).
eu mr
la of
V sn
no ra
it T
al re
er ir
ro uo
C F
de de
izl zi
a la
mr mr
o o
N N
be determined separately and then the phase shift can be Crl (y 0 ) ≡ Ir ⊗ Il (y) = Ir (k)Il (k − y 0 ). (4)
k=0
obtained by performing the subtraction, as indicated in
equation (3). Due to the specific characteristics of the This gives a measure of the similarity of the two functions
present work measurements, it is only necessary to know as one of them is displaced over the other. The position
the phase difference. Consequently, it is easier to calculate of the correlation maximum corresponds to the maximum
the relative fringe displacement and the fringe period and overlapping conditions and, in the case of fringe correlation,
compute φ = 2π 1/P . it directly provides the fringe displacement.
The correlation product of equation (4), normalized to
its maximum value (100), is depicted in figure 3. The exact
3.1. Fringe displacement measurement determination of the position of the maximum is difficult,
because of the noisy ripple present in the signal, which is
A simple method to determine the fringe displacement is to primarily due to the LCTV pixel structure visible in the
perform the one-dimensional correlation product between fringe patterns (figure 2).
766
Complex modulation characterization of liquid crystal devices
where M is the number of averaged peaks (if all of them Figure 6. The correlation product between the cosinusoidal
are considered, M = 2N/P ). The result for the signals adjusted signals.
presented in this work is 1 = 29.5 pixels.
the Fourier transformation process so the total number of
3.2. Period measurement points increases, thus providing better resolution. We have
The period of the fringes can be determined either from considered that εf = 1/64 is an acceptable value for our
the left or right cosinusoidal distributions or from their calculations.
correlation product. We have chosen the latter, which In figure 7, a Fourier plane comparison between the left
directly gives an average measure. The Fourier transform and right fringe distributions and their correlation product
of the signal has to be computed and the frequency that is preserved. Note that while the left and right signals
gives a maximum value determined. From equation (2), an have a 0.25 resolution (3N zero points were added to the
accuracy error εf in the frequency calculation is translated original signal) the correlation signal looks continuous, as
f the number of points is high. The period obtained from the
into a period error, εP ,
Fourier transform correlation function is P = 61.94 pixels,
so the final fringe shift is
∂P P 2
f
εP = εf = εf (6)
∂f N φ = 2π 1/P = 0.95π = 171◦ .
where all the errors are taken to be positive.
As a consequence of the fast Fourier transform routine 3.3. Precision and error discussion
employed for the calculations [11], the resolution error The whole phase shift measurement method involves var-
εf is the ratio between N and the number of points of ious steps with an associated error, as the digitalization
the signal to be transformed. As the correlation product process, the one-dimensional fringe distribution approxima-
has 2N information terms, εf = 0.5, which would give tion, the cosinusoidal adjustment and the finite correlation
f
εP = 16 pixels for a period P = 64. In order to product obtention. All these sources of error should be con-
reduce this error, it is possible to add non-data zeros to sidered, apart from the implicit precision error derived from
767
E Martı́n-Badosa et al
right
correlation
Frequency
Figure 7. Fourier transform of the correlation product and of the single signals.
εφ
0.045
tion, the corresponding perpendicular cosinusoidal distribu- P, f
0.035
the measures, the phase shift obtained when all the LCTV ε∆, f
φ
panel has the reference grey level (gl = 0), which should
ε∆
π
0.030
768
Complex modulation characterization of liquid crystal devices
5. Characterization curves
f
and the right one P − εP , the correlation maxima would
be displaced from their original position by an amount of The amplitude and phase characteristics for specific
f
2εP . This gives polarizers and potentiometers configuration can be jointly
displayed on a complex plane curve (as in figure 10), where
P pix P2 each point stands for a single grey level, from gl = 0
ε1 = ε1 + 2 εf (11)
N N to gl = 255. The modulus of a vector from the origin
of coordinates to the considered point gives the amplitude
and a maximum total error εφmax (P ) = εφ (φ = π, P ), modulation and the angle within the real-positive axis is the
εφmax (P ) P ,dig P ,f 1,f
phase shift.
= εφP + εφ1 = εφ + εφ + εφ1,res + εφ The measurement has been repeated for 16 evenly
π
dig pix spaced grey-level values, corresponding to the squared
εP P ε P
= + εf + 1 + 2 εf (12) symbols in the curves. The remainder of the points (circles)
P N N N are obtained by linear interpolation.
dig pix
with N = 512, εf = 1/64, εP = ε1 = 0.5. Figure 8 We show two interesting LCTV configuration results:
shows the different contributions to the total error and the first one (figure 10(a)) corresponds to a high-contrast
their dependence with P . We can see that the intuitive mode, in which the device transmittance (amplitude square)
behaviour is well reproduced, giving a minimum value for increases with the grey level, reaching a normalized
the periods of about 64 pixels (eight maxima and eight maximum value of 1 for gl = 255. The contrast of the
minima in the digitized picture). The total maximum error modulator, defined as
is in this case 0.02π . Nevertheless, due to the variation
of the LCTV response with time, as a consequence, for C = Tmax /Tmin (13)
example, of the temperature dependence, we have observed
where T stands for the transmittance, is about 126:1. The
a variation of 0.05π over different measures taken in the
orientations of the first polarizer and the analyser (polarizer
same experimental conditions. This is the reason why we
placed after the LCTV) are indicated in degrees relative to
can state that, for a phase modulation characterization, the
the laboratory vertical, being positive if the polarizer turns
deviation from the real phase shift value is about 0.05π
clockwise in the direction of light propagation and negative
(9◦ ).
if it turns counterclockwise. The positions of the brightness
(Bn), contrast (Cn) and colour (Cl) potentiometers available
4. Amplitude modulation method on the videoprojector are also specified.
Figure 10(b) shows a phase-mostly configuration,
With the method described above, it is possible to determine because the transmittance is almost constant (the minimum
the change in light phase as it goes through the LCTV. A transmittance is about 85% of the maximum) and 1.9π
related intensity light variation is always present. For a phase modulation is achieved. The fact that a full
complete device calibration, it is then necessary to know its 2π radian modulation is not attained is due to LCTV
transmittance performance, usually called amplitude light limitations, so that the operation curve found is the best
modulation. Next, we briefly describe the transmittance one we can obtain with this device, considering almost
measurement method. constant transmittance. Furthermore, it does not represent a
769
E Martı́n-Badosa et al
Imag o
LCTV FL CCD
Pol: 6
o
1.0 Anal: -5
(Br,Cn,Cl)=(2,10,10)
0.5
PC
gl=0 gl=255
770
Complex modulation characterization of liquid crystal devices
771
E Martı́n-Badosa et al
772