m
Line 11: Line 11:
 
----
 
----
 
===Solution 1===
 
===Solution 1===
a) <math>\gamma=1</math>
 
  
b)
+
a) <math>E\left[Y_x\right] = \lambda_x</math>
 +
 
 +
b) Because the rate of absorption is proportional to the number of photons and the density of the material, so the attenuation of photons obeys the following equation
 +
 
 
<math>
 
<math>
\left( \begin{array}{c}
+
\frac{d\lambda_x}{dx} = -\mu(x)\lambda_x
X_r\\
+
</math>
Y_r \\
+
c) Solve the differential equation in b), we have
Z_r \end{array} \right)=
+
 
\left( \begin{array}{ccc}
+
a & b & c \\
+
d & e & f \\
+
g & h & i \end{array} \right)
+
\left( \begin{array}{c}
+
1 \\
+
0 \\
+
0 \end{array} \right)
+
\left( \begin{array}{c}
+
a \\
+
d \\
+
g \end{array} \right)</math><br \>
+
So
+
 
<math>
 
<math>
(x_r,y_r)=(\frac{X_r}{X_r+Y_r+Z_r}, \frac{Y_r}{X_r+Y_r+Z_r})=(\frac{a}{a+d+g},\frac{d}{a+d+g})
+
\lambda_x = \lambda_0e^{-\int^x_0\mu(t)dt}
</math> <br \>
+
Similarly
+
<math>
+
(x_g,y_g)=(\frac{b}{b+e+h},\frac{e}{b+e+h})
+
</math>,
+
<math>
+
(x_b,y_b)=(\frac{c}{c+f+i},\frac{f}{c+f+i})
+
 
</math>
 
</math>
  
c) The white point of the device is when the input <math>[R, G, B] = [1, 1, 1]</math>
+
d) So the integral of the density, <math>\int^T_0\mu(x)dx </math> can be written as
 
+
 
<math>
 
<math>
(x_w,y_w)=(\frac{a+b+c}{a+b+c+d+e+f+g+h+i},\frac{d+e+f}{a+b+c+d+e+f+g+h+i})
+
\int^T_0\mu(x)dx = -\log\left(\frac{\lambda_T}{\lambda_0}\right\)
 
</math>
 
</math>
  
d)
 
[[ Image:Pro1_solution1_2015_Aug.jpg]]<br />
 
  
e) We are likely to see quantization artifact in dark region.
+
e) <math>\int^T_ \mu(x)dx \simeq -\log \left\( \frac{Y_T}{Y_0} \right\)</math>
 +
 
 
== Solution 2: ==
 
== Solution 2: ==
  

Revision as of 14:50, 18 May 2017


ECE Ph.D. Qualifying Exam in Communication Networks Signal and Image processing (CS)

August 2014, Problem 1

Problem 1 , 2

Solution 1

a) $ E\left[Y_x\right] = \lambda_x $

b) Because the rate of absorption is proportional to the number of photons and the density of the material, so the attenuation of photons obeys the following equation

$ \frac{d\lambda_x}{dx} = -\mu(x)\lambda_x $ c) Solve the differential equation in b), we have

$ \lambda_x = \lambda_0e^{-\int^x_0\mu(t)dt} $

d) So the integral of the density, $ \int^T_0\mu(x)dx $ can be written as $ \int^T_0\mu(x)dx = -\log\left(\frac{\lambda_T}{\lambda_0}\right\) $


e) $ \int^T_ \mu(x)dx \simeq -\log \left\( \frac{Y_T}{Y_0} \right\) $

Solution 2:

a) $ \frac{R}{255}^\alpha=r_{linear}\\ \Rightarrow \gamma=log_{\frac{R}{255}}{(R^{\alpha})}=\frac{ln{(R^{\alpha})}}{ln{\frac{R}{255}}}=\frac{\alpha{ln{R}}}{ln{R}-ln{255}} $

$ \gamma $ should be 1.

b)

$ P_r= \left( \begin{array}{ccc} a & b & c \\ d & e & f \\ g & h & i \end{array} \right) \left( \begin{array}{ccc} 1 \\ 0 \\ 0 \end{array} \right) = \left( \begin{array}{ccc} a \\ d \\ g \end{array} \right) \\ \Rightarrow x_r=\frac{a}{a+d+g} , y_r=\frac{d}{a+d+g} \\ P_g= \left( \begin{array}{ccc} a & b & c \\ d & e & f \\ g & h & i \end{array} \right) \left( \begin{array}{ccc} 0 \\ 1 \\ 0 \end{array} \right) = \left( \begin{array}{ccc} b \\ e \\ h \end{array} \right) \\ \Rightarrow x_g=\frac{b}{b+e+h} , y_g=\frac{e}{b+e+h} \\ P_b= \left( \begin{array}{ccc} a & b & c \\ d & e & f \\ g & h & i \end{array} \right) \left( \begin{array}{ccc} 0 \\ 0 \\ 1\end{array} \right) = \left( \begin{array}{ccc} c \\ f \\ i \end{array} \right) \\ \Rightarrow x_g=\frac{c}{c+f+i} , y_g=\frac{f}{c+f+i} $

c)

$ W= \left( \begin{array}{ccc} a & b & c \\ d & e & f \\ g & h & i \end{array} \right) \left( \begin{array}{ccc} 1 \\ 1 \\ 1\end{array} \right) = \left( \begin{array}{ccc} a+b+c \\ d+e+f \\ g+h+i \end{array} \right) \\ \Rightarrow x_g=\frac{a+b+c}{a+b+c+d+e+f+g+h+i} , y_g=\frac{d+e+f}{a+b+c+d+e+f+g+h+i} $

d) Pro1 2015 Aug.PNG

e) Gamma correction a quantization will create an effect of dynamic range compression for pixels with small values. This will create dark block of shadings in a gradient region instead of a smooth transition.


Back to QE CS question 1, August 2013

Back to ECE QE page:

Alumni Liaison

Ph.D. 2007, working on developing cool imaging technologies for digital cameras, camera phones, and video surveillance cameras.

Buyue Zhang