All optical image processing using third harmonic generation for image correlation

Image Processing Using THG for Image Correlation Halim

All Optical Image Processing
Using
Third Harmonic Generation
For
Image Correlation

Md. Faisal Halim (Faissal)

Optical Information Processing Course

Prof. Dorsinville

Monday, 20th December, 2010

Monday, 20th December, 2010 1


Table of Contents

Topic Page Number
Paper Reviewed 3
Motive: Computing 3
Optical Image Correlation 4
4-f Architecture 5
Third Harmonic Generation 6
Contrast of the Detected Image 7
Importance of Orientation 8
Time Gating 8
Future Work 10
Conclusion 10
References 11



Paper Reviewed:

Third-harmonic generation and its applications in optical image processing

Canek Fuentes-Hernandez, Gabriel Ramos-Ortiz, Shuo-Yen Tseng, Michael P. Gaja

and Bernard Kippelen

J. Mater. Chem., 2009, 19, 7394–7401

DOI: 10.1039/b905561d

Motive: Computing

Just as an anecdote, an IBM Road Runner Supercomputer (the world’s fastest

supercomputer) occupies 560 m2 of real estate and can run 1.7x1015 Floating Point

Operations Per Second. 2 Dimensional image processing, on the other hand, which

implements matrix convolutions and multiplications, when implemented with ultrafast

nonlinear optical processes, can supersede 1016 Operations Per Second1.

So, by utilizing the massive parallelism offered by 2D signal processing, the high

repetition rates of the ever faster laser sources2, and inherently ultrafast nonlinear optical

processes it will be possible to produce ever faster computers. This will scale up the

speed of computing, and thus extend their range of applications far beyond what the size

reduction of electronic chip components would otherwise allow, and so optical

information processing may become the new avenue that allows the computing industry

to keep up with Moore’s Law.



Optical Image Correlation

Optical image correlation is just one of the many applications of optical computing, and it

may one day make complex tasks that are done after the digitization, where computations

are performed on an electronic readout, a thing of the past. To give an analogy, when a

modern radar system detects signals from a target it takes a digital readout, then

compares the spectral (and temporal) characteristics against a database of known

signatures to identify the target. Such a system has a number of disadvantages, like (a)

digitization is an intermediate step, and (b) digital signal processing, using electronics, is

an inherently slower process than optical processes. If such processing could be done

even before the received signal is read out by the radar system, however, then the system

could perform a lot faster. This is where optical image correlation comes in. Image

correlation, when done in the optical regime, is useful because then physical processes

like harmonic generation can be utilized to generate a flag signal if and only if a target

signal if detected (the flag signal, at a shorter wavelength than the target signal, alerts the

system, or its users, of the presence of the intended target’s signal). Third harmonic

generation (THG), while less efficient than second harmonic generation, is a good choice

of process for optical image correlation because it can happen where three coherent light

beams interact1, much like the interaction of three coherent beams of equal wavelength,

but different propagation vectors, in four wave mixing process where the three incoming

beams sum up to their third harmonic3. In fact, there are groups that are doing four wave

mixing using the 4-f architecture (which is popular for image processing4)5,6.



In many current image correlation systems the incoming signals are sent through a spatial

light modulator (SLM) and the signal that one is looking for, as well as the background,

are at the same wavelength. Using THG, however, the presence of a sought after signal

will trigger THG, so the lack of the third harmonic of the fundamental wavelength will

indicate the absence of the sought after signal (more on this later). This does not remove

the necessity of the SLM, however, as it may still be required to project the spectral

characteristics of a sought after signal, and the signal from the target will have to interact

with it (and also a Gaussian signal) for THG to happen.

4-f Architecture

The 4-f architecture used by this group is as follows:

Figure 1: 4-f Architecture; an SLM can be used in the position of h, so that one may change what

object one wants to detect.



Figure 1 shows that three coherent beams, when propagating non collinearly, can undergo

THG in a suitable material (in this case, a stable, polymeric, solution processable material

with a high χ(3), preferably a high absorption at the fundamental wavelength, and a low

absorption at the third harmonic). If one of the beams goes through a pinhole (r), one

goes through a mask (h – the target being searched for) and the other (g) acts as the signal

comes in from a remote source (rather than through the diffractive optical element) then

the image in the back plane will show bright spots where the spatial features and

orientation of g match the features and orientation in h. In other words, wherever in the

image g there is a ‘copy’ of h, there will be a corresponding bright spot (in the shape of h)

in the image in the back plane.

Third Harmonic Generation

Figure 2: Propagation vectors for THG7

Owing to the direction of the propagation vectors of the incoming light for THG (see

Figures 27 and 3) the incoming light and the THG light are spatial separated, and thus do

not have to be separated using further optics, and there is no ambiguity as to the source of

the light, upon detection at the back plane.



Figure 3: Vector mismatch between the exciting and THG modes seen for (a) two beam excitation

and (b) 3 beam excitation, along with the resulting spatial separation of the beams in the image plane

Contrast of the Detected Image

In such a system the radius of the pinhole r (from Figure 1) dictates the sharpness,

contrast and brightness of the detected image, as shown in Figure 4.

Figure 4: Size of r decides contrast and brightness of detected image



Importance of Orientation

A major drawback of the system used here is that the orientation of the detection mask (h,

from Figure 1) dictates how well (and if) the signal being sought is even detected, as

shown in Figure 5.

Figure 5: Orientation of mask, relative to signal, dictates detection probability – the mask and the

signal have to be oriented the same way for a recognizable THG image to be generated.

Time Gating

Owing to the fact that an ultrashort pulse will get elongated by scattering media in a real

system (see Figure 6) a time gating system will reduce the effects of scattering (Figures 7

and 8), by only doing THG with the ballistic photons.



Figure 6: Realistic System; note that h is used as the signal from the remote location, rather than g,

here, as opposed to the text of this paper. ‘r’ is used as the gating signal for THG. Clearly, the THG

signal is separated, spatially, from the NIR (fundamental) signal.

Figure 7: Effects of time gating: the bottom row shows the delay between the incoming signal and the

gating (mfp = mean free path). The excitation was a 1550nm, and the detection was done at 517nm.

Figure 8: Time gating the system's pulse, from Figure 6, can be used to select only the ballistic
photons from a an incoming pulse for THG.



Future Work

New kinds of molecules, like octupolar molecules, with higher absorption in at the

fundamental wavelength, and lesser absorption at the THG wavelength will enhance such

applications, even though this group has done some work on the materials used to make

the film for THG8,9,10.

Conclusion

An optical image correlator has been implemented, and it is all optical. Complete

independence from the use of SLMs has not yet been achieved, but future all optical

systems could eventually do for optical surveillance what radar target databases has done

for air surveillance, only they would work much faster.



References

1. Third-harmonic generation and its applications in optical image processing

Canek Fuentes-Hernandez, Gabriel Ramos-Ortiz, Shuo-Yen Tseng, Michael P.

Gaja and Bernard Kippelen

J. Mater. Chem., 2009, 19, 7394–7401

DOI: 10.1039/b905561d

2. Recent developments in compact ultrafast lasers

Ursula Keller

Nature 424, 831-838 (14 August 2003)

DOI: 10.1038/nature01938

3. Springer Handbook of Lasers and Optics

Copyright, 2007

Frank Träger (Ed.)

4. Optical Signal Processing

Anthony VanderLugt

Wiley, 2005

5. Degenerate multi-wave mixing inside a 4f imaging system in presence of

nonlinear absorption

K. Fedus, G. Boudebs, H. Leblond

Appl Phys B (2010) 100: 827–831

DOI 10.1007/s00340-010-4004-z

6. Ultrafast nonlinear optical properties of alkyl phthalocyanines investigated using

degenerate four-wave mixing technique



R. Sai Santosh Kumar, S. Venugopal Rao, L. Giribabu, D. Narayana Rao

Optical Materials 31 (2009) 1042–1047

7. Four-Wave Mixing and its Applications

CW Thiel

http://citeseerx.ist.psu.edu/viewdoc/download?

doi=10.1.1.118.63&rep=rep1&type=pdf

20th December, 2010

8. "Ultrafast optical image processing based on third-harmonic generation in organic

thin films“

Fuentes-Hernandez et. al.

APPLIED PHYSICS LETTERS 91, 131110, 2007

9. "Thick Optical-Quality Films of Substituted Polyacetylenes with Large, Ultrafast

Third-Order Nonlinearities and Application to Image Correlation“

Chi, et. al.

Adv. Mater. 2008, 20, 3199–3203

10. “Second-order nonlinear optical properties of octupolar molecules structure–

property relationship”

Kim, et. al.

J. Mater. Chem., 2009, 19, 7402–7409


All optical image processing using third harmonic generation for image correlation

More Related Content

What's hot (20)

Viewers also liked (8)

Similar to All optical image processing using third harmonic generation for image correlation (20)

More from M. Faisal Halim (15)

Recently uploaded (20)

All optical image processing using third harmonic generation for image correlation