Question: Find 3 bit IGS coded image and calculate compression factor and bits per pixel (BPP)
0

Given $F = \begin{bmatrix} \ 13 & 54 & 12 \\ \ 13 & 11 & 57 \\ \ 11 & 10 & 12 \end{bmatrix}$

Mumbai University > Computer Engineering > Sem 7 > Image Processing

Marks: 8 M

Year: Dec 2012

ADD COMMENTlink
modified 19 months ago by gravatar for shirvandkar.prasad shirvandkar.prasad0 written 3.1 years ago by gravatar for Juilee Juilee2.3k
0

(i) To find the 3 bit IGS coded image

Size of i/p image = 9 pixels = 9 bytes

BPP = 6 bits for 3 bit IGS coded image

Therefore, Size= 9 x 6 bytes

enter image description here

IGS coded image:

001 111 001
010 001 111
001 001 010

Size of IGS coded image =27 bits

Compression ratio=50%

(ii) To find decoded image, MSE and PSNR.

To find decoded image:

IGS coded pixel value Decoded pixel in binary Output pixel value
001 001000 8
111 111,000 56
001 001,000 8
010 010,000 16
001 001,000 8
111 111,000 56
001 001,000 8
001 001,000 8
010 010,000 16

Decoded image:

8 56 8
16 8 56
8 8 16

To find MSE (Mean Square Error):

$MSE = \frac{1}{MxN} \sum_{x=0}^{M-1} \sum_{y=0}^{N-1} [f(x,y) - \hat f(x,y)]^2 \\ MSE = \frac{1}{9} [(5)^2+(-2)^2+(4)^2+(-3)^2+(3)^2+(1)^2+(3)^2+(2)^2+(-4)^2] \\ MSE= \frac{1}{9} [25+4+16+9+9+1+9+4+16] \\ MSE= 10.33$

To find SNR (Signal to Noise Ratio):

$SNR = \frac{\sum_{x=0}^{M-1} \sum_{y=0}^{N-1} [f(x,y)]^2}{\sum_{x=0}^{M-1} \sum_{y=0}^{N-1} [f(x,y) - \hat f(x,y)]〗^2} \\ Signal = [(13)^2+(54)^2+(12)^2+(13)^2+(11)^2+(57)^2+(11)^2+(10)^2+(12)^2] \\ Signal = 7133 \\ Noise = [(5)^2+(-2)^2+(4)^2+(-3)^2+(3)^2+(1)^2+(3)^2+(2)^2+(-4)^2] \\ Noise = 93 \\ SNR = \frac{7133}{93} = 76.698$

To find PSNR (Peak Signal to Noise Ratio):

$PSNR = \frac{(Peak signal value)^2}{MSE} \\ For 6 bit image Peak value = [111 111] = 63 \\ PSNR = \frac{(63)^2}{10.33} = 384.22$

ADD COMMENTlink
modified 3.1 years ago  • written 3.1 years ago by gravatar for Juilee Juilee2.3k
0

How did you get size of igs coded image as 27 bits?

ADD COMMENTlink
written 19 months ago by gravatar for shirvandkar.prasad shirvandkar.prasad0
Please log in to add an answer.