You are on page 1of 6

High Resolution Spatial Intensity Control using Adaptive Optics

MD. Tashin Ahammad (2017132042)1, MD. Jobair Hasan (2015132005)1,

Irfan Hossain (2015135054)2, Ahmad Al - Imtiaz (2017132048)1


1
Department of Physics, Shahjalal University of Science and Technology
2
Department of Geography and Environment, Shahjalal University of Science and Technology

Abstract
Spatial intensity control can be achieved by passing a homogeneous beam through a refracting sheet of
specific relief function on one of the broad sides, Inverse Laplacian approximation of the relief function
has been gained by utilizing iterative method, Afterward, the solution relief function has been verified
by simulating Snell’s law and producing the associated intensity function projection. Possible adaptive
implementation with Piezoelectric Material-Indium Tin oxide Grid has been discussed.

Keywords: Laplacian Magic window, Adaptive Optics, Piezoelectric Material, Indium Tin oxide.
Introduction:
Magic windows are a 2017 creation with origins in so-called magic mirrors dating back hundreds of
years. Magic mirrors are similar to magic windows in that they can operate as optical Laplace operators.
Parallel light strikes a glass sheet with a smooth surface on one side and a moderate surface variation on
the other. The light is refracted by the magical window and falls onto a wall. Although the bumpiness of
the glass has no discernible shape or pattern, the light density fluctuations on the wall reveal a distinct
image. We must solve a differential equation (Poisson equation) for our desired image in order to
determine the form of the glass surface.
For the desired image, a specific glass surface is required. As a result, we can no longer use a single
glass for several photos, which raises our expenditures. That is why we have offered a method of
implementing the system in such a way that the refractive material can be reused for several images.

Theory [1] & Methods [2]:


The window has a refractive index of 𝑛 and surface relief of ℎ(𝑟) above points in the window plane with
𝑟 = {𝑥, 𝑦} coordinates (figure 1). It is simple to compute the coordinates 𝑅 = {𝑋, 𝑌} of refracted rays in
an image plane at height 𝑧, using Snel's law, in terms of the transverse gradient ∇ℎ = {𝜕𝑥 ℎ(𝑟), 𝜕𝑦 ℎ(𝑟)}:

Figure 1. Refraction by a magic window: geometry and notation

This is the Laplacian image, which is formed by the Laplacian magic window:
𝐼𝐿𝑎𝑝 (𝑅) = 1 − (𝑛 − 1)𝑧∇2 ℎ(𝑅) … (1)

This equation, known as Poisson's equation, is used to find the surface ℎ(𝑅) that will yield the desired
image 𝐼𝐿𝑎𝑝 (𝑅).
𝐼𝐿𝑎𝑝 (𝑅) − 1
∇2 ℎ(𝑅) = − … (2)
(𝑛 − 1)𝑧
To verify that the negative sign is true, observe that ∇2 ℎ(𝑅) < 0 corresponds to local convexity, which
enhances the intensity by focusing light for refractive indices n>1. We have solved this Poisson equation
for ℎ(𝑅) using standard five-point formula of finite difference method with arbitrary boundary condition.
1 𝐼𝐿𝑎𝑝 (𝑅) − 1
ℎ𝑖,𝑗 = (ℎ𝑖+1,𝑗 + ℎ𝑖−1,𝑗 + ℎ𝑖,𝑗+1 + ℎ𝑖,𝑗−1 + … (3)
4 (𝑛 − 1)𝑧
Afterward, we iterated the initial solution by applying Gauss-Seidel method and reach our final solution.

Physical Implementations and Possible Issues:


So far, the theoretical aspects of the idea have been discussed and verified. As a physical problem there
are several ways to implement the idea physically and each implementation has its own advantages and
disadvantages. There are two major categories of implementations of the idea. Namely, the static
implementations and the adaptive or dynamic implementations.
The static Implementations: The whole physical arrangement of the problem is solved for a particular
intensity function and the refracting medium is pretty much static (i.e., a precast transparent sheet with
a specific relief map.). The physical stability in this implementation is excellent as there is no moving
part. The major disadvantage is that there is no way of fast adaption of the system for a new intensity
function. It can serve the purpose where there is a need of greater stability and reliability with almost no
adaptation.
The adaptive implementations: The whole idea is physically implemented with an adjustment
mechanism for controlling and setting the refracting medium for a particular intensity function. The
whole system becomes more versatile as any intensity function can be achieved almost instantaneously.
The cost of this greater fluidity is that the system becomes much complex with many moving parts which
increases the chance of failure. A sheet of transparent material with fixed index of refraction can be
shaped into the desired shape consisting of the relief map associated with a particular intensity function.
CNC or other high-precision milling techniques can be used to grind out a specific surface-height
function.

Figure 2. Schematic diagram of Piezoelectric-ITO adaptive implementation


The adaptive implementation is filled with much more possibilities together with much more pitfalls.
The refracting medium in this case has to be of a material which can be manipulated with high precision
by a physically-feasible mechanism. The are many materials with this physical feature but each has to
be tested in the light of precision and reliability. Regular fluids with viscosities around water cannot be
used as too much fluidity comes with harder control challenges. Ultrasonic standing waves could be used
to hold the upper surface up to a certain height function, but the high precision requirement of the height
function makes the production of such waves very challenging. Semi-fluids or gels with higher
viscosities can ease the challenge. As of now, the most promising implementations seem to be using
materials with reactionary properties to the applied electric or magnetic fields. The materials which
physically deform in the presence of an electric field or the so-called 'Piezoelectric materials' can be used
as the refracting medium. A transparent sheet made of quartz can be manipulated to a specific shape by
sandwiching it between two array of individually addressable Indium-Tin Oxide (ITO) electrodes
arranged in a rectangular grid. The electric field between the ITO arrays can be controlled by adding or
removing electric charges from the individual electrodes. The usage of transparent electrode materials
like ITO is inevitable as the whole apparatus has to be transparent. One of the broader surfaces of the
sheet must be mechanically fixed to keep it flat while the other surface should be deformed.
Both the static and the adaptive implementations of the idea come with several promising fields of
implementations. Any physical system where there is a need of high-resolution optical intensity control
can utilize the idea. There are many possible usages in the fields of photolithography, astronomy, and
material science.

Results & Discussions:


With moderate-precision milling techniques and piezoelectric materials the cases of static and adaptive
implementation can be physically achieved. In the above section the verification of physical feasibility
has been shown. We have solved the whole physical problem for a specific intensity function (a 300*300
image) and got relief height function. As for an experiment we took refractive index of the material 1.5
and the distance from glass to screen is 1 unit to make the scaling of the height easier. Afterward, we
verified our result by reversing the whole method and getting the starting intensity function. Here are the
findings for 10, 100, 300 iterations. If We increase the iteration number, we could get actual or more
precise height function and image. But we couldn’t for computational limitations (time-consuming).

(a) (b) (c)


(d) (e)

(f) (g)
Figure 3. (a) Gray scaled original image (b) Height function of the glass (solving Poisson’s Equation by 10
iterations) (c) Projected image by applying Snell's law in height function of (b) (d) Height function of the glass
(solving Poisson’s Equation by 100 iterations) (e) Projected image by applying Snell's law in height function of
(d) (f) Height function of the glass (solving Poisson’s Equation by 300 iterations) (g) Projected image by applying
Snell's law in height function of (f)

References:
1. Berry, M. V. (2017). Laplacian magic windows. Journal of Optics (United Kingdom), 19(6),
06LT01. https://doi.org/10.1088/2040-8986/aa6c4e
2. How Laplace Would Hide a Goat: The New Science of Magic Windows
Appendix (python Code):
from PIL import Image
import numpy as np
import matplotlib.pyplot as plt
import scipy.ndimage.filters

#importing image and coverting it to gray scale image


Img = Image.open('D:/Study/Academics/4-1/Compu II/Project/photo.jpg').convert("L")

#converting the image into array


img_data = np.asarray(img)
plt.imshow(img_data, cmap='gray', vmin=0, vmax=255)
plt.xlim([0,300])
plt.ylim([0,300])
plt.show()

x, y = np.shape(img_data)
z = 1; n = 1.5
temp_data = (- img_data[:][:]) / ((n-1)*z) #right side of equation(2)
[X, Y] = np.meshgrid(np.arange(0, x+2, 1), np.arange(0, y+2, 1))

# Boundary value setting


temp_arr = np.zeros((x+2, y+2))
for i in range(x+2):
temp_arr[i][0] = z/2
temp_arr[i][-1] = z/2
for j in range(y+2):
temp_arr[0][j] = z/2
temp_arr[-1][j] = z/2

# Standard 5-point fourmula of PDE solving


for k in range(10):
for i in range(x):
for j in range(y):
temp_arr[i+1][j+1] =
0.25*(temp_arr[i][j+1]+temp_arr[i+2][j+1]+temp_arr[i+1][j]+temp_arr[i+1][j+2] -
temp_data[i][j])
print("iteration", k, "for index", i ,j)

# Ploting genarated glass


temp_arr0 = temp_arr/np.sum(temp_arr)
temp_arr1 = z-temp_arr0
fig = plt.figure()
ax = plt.axes(projection='3d')
ax.contour3D(X, Y, temp_arr1, 100, cmap='binary')
plt.xlim([300,0]); plt.ylim([0,300])
ax.view_init(-125, 85)
plt.title("h(r) for z= {}".format(z))

# Projected figure
projection = (1 - ((n-1)*z*(scipy.ndimage.filters.laplace(temp_arr))))
fig = plt.figure()
plt.imshow(projection, cmap='gray', vmin=0, vmax=255)
plt.xlim([0,300]); plt.ylim([0,300])
plt.show()

You might also like