Optical image subtraction in fluorescein-doped boric acid glass

Optical image subtraction in fluorescein-doped

boric acid glass S. A. Boothroyd,

L. Chan, P. H. Beckwith,

and J. Chrostowski

The interference of two coherent images with a controlled phase difference between them is shown by using four-wave mixing in fluorescein-dye-dopedboric acid glass and a liquid-crystal television spatial light modulator. We present results showing the digital optical exclusive OR operation with milliwatt optical power and an output that is compatible with CCD camera sensitivities.

1.

Introduction

Image subtraction, in which two coherent images with a relative phase difference of half a wavelength and a similar amplitude variation are overlapped, represents a powerful processing operation in optics. Many areas of application have been suggested for image subtraction, particularly those in which small changes in some object must be manifest. Optical subtraction of one image from another to detect the difference between the two images has been achieved by using interferometric techniques with transparencies that represent the images in separate optical paths.' The use of phase-conjugate mirrors in a Michelson interferometer permits a phase difference of Trto be simply established between the two images at the interferometer output. 2 4 Furthermore the output is stable, being insensitive to the normal interferometer restrictions imposed by vibrations and air turbulence. Although phase-conjugate interferometry (PCI) has many advantages it is still necessary that the two image transparencies be carefully aligned and imaged so that they are correctly overlapped at the output. We propose and demonstrate an approach to image subtraction in which a single spatial light modulator (SLM) is used to input both images in a four-wave-mixing phase-conjugation scheme that encompasses the advantages of PCI while minimizing the necessary optical alignment, and providing for an easily updatable input. The method involves creating two real-time holograms representing different images that have a controlled phase difference between them. The two images are written alternately on a laser beam by transmission The authors are with National Research Council, Institute for Information Technology, Ottawa, Ontario KIA OR6, Canada. Received 28 March 1991. 4004

APPLIED OPTICS / Vol. 31, No. 20 / 10 July 1992

through a SLM, and a phase difference is established between them by reflection from a movable mirror. The piezoelectric mirror is stepped through an interferometric distance in synchronization with a change in the SLM. In this way two gratings can be established in the nonlinear medium, and the phaseconjugate readout of the resultant grating structure is the difference between the two input images when the phase difference between them is rr. 2. Nonlinear Optics of Fluorescein-Doped Boric Acid Glass

Dye-doped glasses and dye film structures have been widely studied as nonlinear materials.i' 4 The nonlinearity arises from saturated absorption. Excitation from the dye singlet ground state to higher singlet states can transfer to the lowest-lying triplet state, which decays nonradiatively. The lifetime of the triplet state is long when the dye molecule is fixed in a rigid matrix which leads to a low saturation intensity and large nonlinearity at low laser powers. The optical constants of the dye glass are determined by the population in the singlet ground state and in the lowest triplet state. Spatial variation in the singlet-triplet occupation gives modulation in the refractive index and the absorption coefficient, which can diffract light to produce phase-conjugate beams. PCI has been demonstrated by using four-wave mixing in films of eosin-doped gelatin and by recording a real-time grating and a permanent grating to achieve double-exposure interferometry. 6' 7 The nonlinear optical characteristics of the fluorescein-doped boric acid sample used in this work have been studied previously in two-wave mixing.8 The sample was 60 Vumthick, and its ground-state absorption characteristics are shown in Fig. 1. We have measured the phase-conjugate reflectivity of this sample at three argon-ion wavelengths, of 457.9, 465.8, and 476.5

E

.600 0

a) 400

0

0200

0U) 350

450 Wavelength (nm)

550

Fig. 1. Ground-state absorption coefficientmeasured for a 60-jimthick sample of fluorescein-dopedboric acid glass.

nm, which were chosen because they fall within the absorption band of the sample that gives resonant enhancement of the nonlinearity. 9 The nonlinear transmission characteristics of the sample are illustrated in Fig. 2. Saturation in the absorption at 457.9 nm is clearly evident as the light intensity is increased, although at 465.8 nm the absorption cross section for the long-lived triplet level is approaching the value for the ground singlet state and at 476.5 nm is greater than the ground-state value, which leads to the absorption increasing with intensity. By following Hercher' 5 we can estimate the ratio between the cross section for absorption from the singlet ground state o and the long-lived triplet state qe from Fig. 2. At 457.9 nm qeI1qo= 0.6, at 465.8 nm uelo =0.8, and at 476.5 nm oel = 2.8. The concentration of fluorescein in our sample was 1018 cm-3 and we have observed the influence of this wavelength variation in the ratio of the cross sections in our two-wave mixing studies. These measurements are important since the maximum steady-state phase-conjugate reflectivity is known to occur near the saturation intensity in these materials. 3 14"16 I ..

The four-wave mixing arrangement shown in Fig. 3 was used to investigate the phase-conjugate reflectivity of fluorescein-doped boric acid glass. Light from the argon-ion laser was divided into three beams that overlapped coherently in the dye-doped glass sample. The two counterpropagating pump beams were 20mW each, the probe beam was 0.3 mW, and the intensity at the sample from the pump beams was 200 mW/cm2 . At 457.9 nm uqoL= 1.69 and the phase-conjugate reflectivity R was measured as 0.28%, at 465.8 nm ao 0L = 0.64 and R = 0.31%, and at 476.5 nm a0L = 0.16 and R = 0.16%; acois the ground state absorption coefficient and L is the sample length. The movable piezoelectric mirror shown in Fig. 3 was stationary throughout these measurements. The three beams were all vertically polarized at the sample; however, we also detected a phase-conjugate signal when the image beam was cross polarized relative to the pump beams.7"10 The reason for such unexpected behavior arises because the randomly oriented fluorescein molecules are fixed in the glass structure and some experience the electric-field variation of both the cross-polarized pump and probe beams. This polarization-dependent absorption of the fluorescein molecules can also be used to induce dichroism and birefringence and may be exploited for image processing." The rise time for gratings that are written in saturable absorbers depends on the pump intensity and the saturation value.13,'4"16 The rise time decreases with an increasing ratio of pumpto-saturation intensity,14 and gratings can be created much faster than they decay.5 The four-wave mixing gratings written in the dye glass decay with a time constant that is the triplet-level lifetime, and in fluorescein-doped boric acid glass is of the order of 0.1 s (see Ref. 9) to 1 s (see Ref. 5), and depends on

temperatures was - 300 ms. 3.

In our work the grating decay time

Characterization of the Spatial Light Modulator

The SLM used in this work was a Panasonic pocket color television, Model PC-3T20. Parallel polarizers were removed from the front and back of the liquidcrystal display. The display is composed of a raster-

476.5 nm

.4 .,............#±... .......... i 801 I .. *i"" *~...

o

e0

.-

601-

465.8 nm _

-...........

UO

__

.E U, C:

(a

A.. CGD

401 -

A

air

-__

457.9 nm

II

Il

IllIllI

I 1

- _

Il

II..ll 10

___-

dopedglass

PZT

201 - I U.0.1

-4---

--A" A'

I

I

I 111 .

._P

I11

100

1000

10000

Intensity, mW/cm 2

Fig. 2. Intensity-dependent transmission of fluorescein-doped boric acid glass. A,-, and * are the measured data points at 457.9, 465.8, and 476.5 nm, respectively.

Fig. 3. Experimental arrangement: PC, personal computer.

PZT; piezoelectric mirror;

10 July 1992 / Vol. 31, No. 20 / APPLIED OPTICS

4005

scanned array of 240 x 372 pixel cells, each covered

RS-170(NTSC)

by either a red, blue, or green filter, as shown in figure 4. Both the red and green pixels prevented transmission at 465.8 nm, which is the wavelength of maximum phase-conjugate reflectivity from the fluorescein-doped boric acid glass used in a demonstration of PCI. The unit accepts the standard RS-170 video signal from either a television camera or is produced by a personal computer. The liquid-crystal display is positioned, as shown in Fig. 3, between polarizing elements and with its long side at right angles to the plane of the diagram. The television display is normally operated between parallel polarizing elements; when the television is off each pixel rotates the polarization of the incident light. We found, however, that in this case the contrast between transmitting and blocking pixels was not as high as when the device was between cross-polarizing elements. A p-polarized incident beam was ensured by a double Fresnel rhombus, and a polarizer at the output of the SLM permitted transmission of s-polarized light. The transmission of an area of 6 x 4 blue pixels was measured in the "on" (no light transmission) and "off" (transmitting) states, giving a contrast ratio of 1:40. The overall transmission of the SLM polarizer combination in the blue was 5%. The input video signal to the SLM is 2:1, interlaced at 30 Hz. A field consisting of half the total number of scan lines is displayed in 1/60 s and the second field in the next 1/60 s; thus it takes 1/30 s for a completely new pattern to be displayed on the SLM. In our application we synchronize the piezomirror phase shift with the appearance of each pattern on the beam transluminating the SLM. Figure 5 shows the relationship between the two patterns A and B, which were input to the SLM, and the stepping of the piezoelectric mirror, which was triggered from the vertical scan synchronization signal. We registered the mirror stepping either in phase with the two fields making up each pattern, as in case 1 in Fig. 5, or in antiphase with the two patterns, as in case 2 in Fig. 5. Figure 6 shows the light transmission measured through a region of 6 x 4 blue pixels for three

Video Signal

Piezo Drive Voltage

Pattern A I Pattern B I Pattern A I

|

Field1 Field2 Field1 Field2 Field1 Field2

r#1 L #2

Vertical SyncSignal

14

1/30 s

Synchronization of the alternating television signal representing patterns A and B with the stepping of the piezoelectric mirror. NTSC, National Television Systems Committee. Fig. 5.

different positions on the SLM: the start of the raster scan; the middle of the display area; and the end of the raster scan. The figure shows the transmission for pixels that were transmitting in pattern A and blocking in pattern B and the transmission for an adjacaent area of pixels that were blocking in A and transmitting in B. Figure 6(a) shows that when the piezomirror is stepped in phase with the two patterns the transmission of the pixels is synchronous with the programmed phase shift over the whole area of the display. Figure 6(b) shows that, when the mirror stepping is out of phase with the pattern alteration, image subtraction could not be achieved in the scheme described above because the phase shift on the probe beam occurs in the middle of each light pattern. The Panasonic SLM incorporates active-matrix electrodes so the state imposed on each pixel does not decay

throughout the field period. When an area of pixels was transmitting in both patterns no modulation in

the transmitted light was detected as the patterns were alternated. In our experiment the two fields that make up A and B were identical and the SLM, with only half the number of lines as the television input signal, was switched fully in one field period. Our measurements have shown that suitable timing between the light modulation and phase shift can be achieved over the whole display, and we have obtained image subtraction by using all areas of the television screen. 4.

Phase Conjugate Interferometry in

Fluorescein-Doped Boric Acid Glass

-

1 mm Fig. 4. Liquid-crystal television pixel scale and arrangement: red filter; G, green filter; B, blue filter. 4006

APPLIED OPTICS / Vol. 31, No. 20 / 10 July 1992

R,

By using the arrangement shown in Fig. 3 we have investigated PCI in fluorescein-dye-doped glass by modulating the image beam with a SLM. The piezoelectric mirror and the SLM were under computer control and a square-wave voltage signal enabled the mirror to be moved between two positions, giving a known phase shift to the image beam. In this way two transmission patterns could be alternated on the SLM at video rate and two gratings could be established in the nonlinear medium. When a pixel permits transmission in both patterns with a relative path difference of half a period the index variations

0

20

40 60 Time (ms) (a)

80

100

0

20

40 60 Time (ms) (b)

80

100

Fig. 6. Timing of laser light transmission through the SLM and polarizer with the stepping of the piezoelectric mirror. (a) The alternation of patterns A and B was synchronized with the mirror displacement, case 1 in Fig. 5; (b) the mirror stepping was out of phase with the changing pattern, case 2 in Fig. 5. The top group of traces in both (a) and (b) shows the light signal detected through a region of pixels that were transmitting in pattern A and opaque in pattern B. The middle group shows the light signal detected through a region of pixels that were blocking in pattern A and transmitting in pattern B. The dashed curves denote a region of pixels at the start of the television scan, the dotted curves denote a region in the middle of the television screen, and the solid curves denote a region of pixels at the end of the television scan. The bottom trace in each figure represents the voltage signal to the piezomirror.

resulting from the two patterns are complementary and the time-averaged material response is zero. For a pixel transmitting throughout one pattern only, the grating is being reinforced many times within the material response time and a steady-state output is produced. However, we noticed that the time constant for fluorescein-doped boric acid glass was slightly too long to counter fully the effects of mechanical vibration in our PCI. At the input to the nonlinear medium the two field amplitudes arising from the alternate SLM transmission states may be expressed as EA

= TA expi(OLM

- kA

EB = TB eXPi(kSLM +

r),

where the real amplitude transmission of the SLM for the two patterns is TA and TB and the amplitude of the beam incident on the SLM has been taken as unity. kA and kB are the wave vectors for the two image beams and r represents spatial coordinates measured from the SLM. is the extra phase difference between TA and TB produced by the movable mirror and ISLM represents any static phase irregularity introduced across the wave front after transmission through the SLM. Following four-wave mixing in the dye glass, the generated phase-conjugate beam that retraces the path back from the nonlinear interaction to the SLM is EA*

=

rATA expi&4 LM + kA

EB*= rBTB expi(-si

-

2 = I = EA + EB*1

+ kB.

where taking the complex conjugate is indicated by the asterisk. As described in Fig. 3 a portion of this beam is redirected before reaching the SLM, and is recorded at the SLM phase-conjugate image plane,

rA2TA2

+ rB2TB2 + 2 rArBTATB

cos(,).

The complex phase-conjugate reflectivities rA and rB can be expressed as intensity reflectivities RA and RB: EA*(0) 2 RA = IrA 12 = E(0)

RB

r), kB

where the intensity is

=

rB 12

=

EB (0)

The reflectivities are defined at the input to the nonlinear medium, as indicated by the zero in parentheses, which is given for the field amplitudes. The conditions of phase-conjugating medium and pumpbeam intensities in the four-wave mixing process are the same for both input beams A and B, and we can expect the phase-conjugate reflectivities for the two inputs to be the same. Abrams and Lind1 7 have analyzed four-wave mixing in a saturable absorber, and they give expressions for the reflectivity in which the imaginary part depends on only the laser frequency detuning from the material absorption line center, which is constant in our experiment, provided the pump beams in the four-wave mixing process are phase conjugate. The real part of the reflectivity determined for a two-level system depends on material parameters such as the saturation intensity and the product of the pump-beam amplitudes, which were also constant throughout our work. Fluorescein-dye glass has been shown to be well approximated as a two-level system8 9 and although a full analysis for the phase-conjugate reflectivity in the case of a modulated input image is not given we 10 July 1992 / Vol. 31, No. 20 / APPLIED OPTICS

4007

consider it reasonable as a first-order approximation to set rA = rB. In this case the intensity at the phase-conjugate image plane is ITA + TB exp(i4) 2 and, setting + = rr, we have amplitude image subtraction i.e., ITA - TBI2. When TA and TB represent digital 1 and 0 transmissions, i.e., light on and light off pixel states, the digital exclusive-ORoperation results, which = is shown experimentally in this work. When 2rr/3 or

cos(0 =

+

= 4rr/3 the

- '/2

OR

function can be achieved as

and the intensity at the phase-conjugate

image plane is proportional

to (TA2 + TB2

-TATB).

The two images are introduced in the processing system by the same SLM and are, therefore, automatically registered at the phase-conjugate image plane.

This has other advantages in addition to automatic alignment for the interference of coherent images. Two identical images may be input, although displaced from each other by one or more pixels, which leads to a phase-conjugate output that gives a spatial derivative operation in the direction of the image translation, i.e., edge enhancement.' 8 Figure 7 shows a series of photographs in which image subtraction is demonstrated. Figures 7(a) and 7(b) are the phase-conjugate outputs when a small area of the SLM, representing patterns A and B, respectively, was illuminated by the probe beam. The pattern on the SLM was constant while these images were produced and individual pixels were

(a)

(b)

(c)

(d)

Fig. 7. Phase-conjugate image recorded by the CCD camera shown in Fig. 3. (a) Phase-conjugate output when pattern A is displayed on the SLM, (b) output for pattern B, (c) phase-conjugate output that results when both A and B are alternated on the SLM at video rate, (d) patterns A and B are alternating on the SLM and have a ir phase shift between them. 4008

APPLIED OPTICS / Vol. 31, No. 20 / 10 July 1992

easily resolved in the output. When these two patterns were alternated on the SLM, as described above, but without any phase shift from the piezomirror, then the phase-conjugate image in Fig. 7(c) was recorded. When the phase shift was set to 7r the output represents image subtraction or, in this case of on and off pixels, the exclusive OR operation [Fig. 7(d)]. Since both input images follow the same optical path any stationary phase irregularity on the optical components or on the SLM is accounted for in the gratings from both input patterns, and when the phase difference between them is Trdestructive interference takes place and the cancellation is complete. Thus the optical quality of the SLM is not a factor in this approach. By stepping the piezoelectric mirror to produce a phase shift of 2'T/ 3 we have also observed a phase-conjugate output that represents the optical logic OR operation, although this is not reported here. By using our technique, in which the interference between two or more images is effected within a nonlinear medium, we have recently demonstrated 18 both the OR operation and image subtraction over the whole area of the SLM, although, in this case, the nonlinear medium was photorefractive barium titanate. Conclusions

We have described a simple scheme in which fourwave mixing phase conjugation in a dye-doped glass operating with only a milliwatt power laser has been combined with an inexpensive liquid-crystal television SLM to demonstrate digital-optical image subtraction. Although the present SLM has gray-scale capability, accurate analog image subtraction over the whole display area is complicated by the need to synchronize the rr phase shift with each pixel when the two images are switched at video rate. This situation is alleviated somewhat by the fact that light modulation occurs over one, not two, field periods when the two fields making up an image are the same. By using a controllable piezoelectric mirror to introduce a fixed phase shift over the whole image and a nonlinear medium in which a time-averaged grating is developed we have access to both the relative strengths of the gratings produced by the two images through their exposure time and to the relative phase difference between them. The fluorescein-doped glass used in this work was kindly provided by Wayne Tompkin and Robert Boyd of The Institute of Optics, University of Rochester, Rochester, New York. References 1. C. P. Grover and R. Tremblay, "Real-time image subtraction

using complementary photographic diffusers," Appl. Opt. 21, 2666-2668 (1982).

2. S. K. Kwong, G. A. Rakuljic, and A. Yariv, "Real time image

subtraction and exclusive OR operation using a self-pumped phase conjugate mirror," Appl. Phys. Lett. 48,201-203 (1986). 3. A. E. Chiou, and P. Yeh, "Parallel image subtraction using a phase-conjugate Michelson interferometer," Opt. Lett. 11, 306-308 (1986).

4. S. A. Boothroyd, and J. Chrostowski, "Interferometer," U.S. patent 5,080,466 (14 January

1992).

5. T. A. Shankoff, "Recording holograms in luminescent materials,"

Appl. Opt. 8, 2282-2284 (1969).

6. K. Nakagawa, and H. Fujiwara, "Real-time and doubleexposure phase conjugate interferometries using eosin-doped gelatin film," Opt. Commun. 70, 73-76 (1989). 7. H. Fujiwara, K. Nakagawa, and T. Suzuki, "Real-time image

subtraction and addition using two cross-polarized phase conjugate waves," Opt. Commun. 79, 6-10 (1990).

8. S. A. Boothroyd, J. Chrostowski, and M. S. O'Sullivan, "Determination of the phase of the complex nonlinear refractive index by transient two-wave mixing in saturable absorbers," Opt. Lett. 14, 948-950 (1989). 9. M. A. Kramer, W. R. Tompkin, and R. W. Boyd, "Nonlinearoptical interactions in fluorescein-doped boric acid glass," Phys. Rev. A 34, 2026-2031 (1986). 10. W. R. Tompkin, M. S. Malcuit, R. W. Boyd, and J. E. Sipe,

"Polarization properties of phase conjugation by degenerate four-wave mixing in a medium of rigidly held dye molecules," J. Opt. Soc. Am. B 6, 757-760 (1989).

11. M. S. O'Sullivan, and P. Myslinski, "Joint transform correlator based on photo-induced anisotropy in fluorescein-doped boric acid glass," in Spatial Light Modulators, Vol. 14 of OSA

1990 Technical Digest Series (Optical Society of America, Washington, D.C., 1990), pp. 43-46. 12. W. R. Tompkin, M. S. Malcuit, and R. W. Boyd, "Enhancement

of the nonlinear optical properties of fluorescein doped boricacid glass through cooling," Appl. Opt. 29, 3921-3926 (1990). 13. Y. Silberberg, and I. Bar-Joseph, "Transient effects in degener-

ate four-wave mixing in saturable absorbers," IEEE J. Quantum. Electron. QE-17, 1967-1970 (1981). 14. H. Fujiwara, and K. Nakagawa, "Transient phase conjugation by degenerate four-wave mixing in saturable dyes," J. Opt. Soc. Am. B 4, 121-128 (1987).

15. M. Hercher, "An analysis of saturable absorbers," Appl. Opt. 6, 947-954 (1967). 16. S. A. Boothroyd, and C. Grey Morgan, "Temporal development

of phase conjugation in ruby by degenerate four-wave mixing," J. Phys. D 16, L165-L168 (1983). 17. R. L. Abrams and R. C. Lind, "Degenerate

four-wave mixing in

absorbing media," Opt. Lett. 2, 94-96 (1978); errata 3, 205 (1978).

18. S. A. Boothroyd, P. H. Beckwith, L. Chan, and J. Chrostowski, "Multiple grating optical processing in barium titanate," in Photorefractive Materials, Effects, and Devices, Vol. 14 of OSA 1991 Technical Digest Series (Optical Society of America, Washington, D.C., 1991), pp. 423-426; "Optical processing in photorefractive media," in Proceedings of the Canadian Conference on Electrical and Computer Engineering (Canadian Society for Electrical and Computer Engineering, Montr6al, 1991), pp. 52.5.1-52.5.4.

10 July 1992 / Vol. 31, No. 20 / APPLIED OPTICS

4009