06 Both

  • November 2019
  • PDF

This document was uploaded by user and they confirmed that they have the permission to share it. If you are author or own the copyright of this book, please report to us by using this DMCA report form. Report DMCA


Overview

Download & View 06 Both as PDF for free.

More details

  • Words: 1,263
  • Pages: 4
_.------~ [

REY IS ED

caJRSE:

00·:,.~~

A:W1938-06. ( 3 Houcs

N.ll,

(I)

question

( Total Marks : 100

)

is compulsory

..... ; .the remaining sixquestiom: .. ' I assume suitable additional data if required ,state & justif;! the assumptions ma?,~:. figures to the right indicate marks ... 110.1

; (2) . attempt any 4 questiollsfrom (3) (4) Q.1. on,

State TRUE or FALSE and justify (Answer any ~EVEN) a) Histogram·is a Unique Representation of an image ..

"

,

~

the imllgc ...

T

l) F F

g) h)

a)

I 14

po

Jf6

., 1--'-

i)

:nd Floor,' Iway Statld 13

I

.I

(04)

., 4 98 713 12 10 5'10 12 6.' 10 J.,512 '4 15 7

Prepare the histogram oflhe image. Perform Histogram Equalization on the given image. (12} For continuous image Histogram Ca,IJbe perfectiy equal!ze~ but it may not be so for the digital image. justify. (04)

ii)

: Classe i. B. Rood; 2314802 :

,

The princi pal fu net ion ot' !TIedion filter is to forel.: pt1'I1lSwith distinct intensities to be more like their neighbors. All ~mngocompressioll teelllliqllcs aro invertible.'. Poorly illuminated imogc.:s call be cosily scgmcn:cd. Computation of discrete convolution in transfo:rn domain is more efficient (hlln ill spatial domain.

What information can be obtained from Histogram of an image? A 16 level image is given below:

b) 18 is

.-

range ?f intensity levels human visual system can ,di)crimi~at~ . sImultaneously IS rather small compared to the total adapjaU9J} range.] c) :' I· Second pass of Histogram Equalization will produce exactly (1e same result·. . . as the tirst pass. ' .... , .. d) r:= Reduction in gray levels produces 'blocky effect' (che;kelboard patterrl) in C I

/'

l,

b)T· :~hetotal

.

1ge .

(}.2.

[il)

..•.

F

n

Q.3. a) Explain operation and application Give3x3 mask wherever applicable: i) low pass filter ii) median filter iii)

I

of each of the following. (iv)

(08] horizontaLlil)£' ,',::ection , .

Laplacian . ""

b) Given below is SxS image. Operate on the central 3x3 pixels by low pass masks & obtain 3x3 images as outputs .. 12

12

Jl

I 4 ~I 2 _ I 3

]0

9-1

6 10

5 15

4

10

:

& high pm tilter ,

61

g~.;--~~ ~_.I~ U:;ing these outputs verify original image= lowpass output discrepancy explain the reasons.

+ Highp:J.ss output.

In case of

(12] [ '!URN

OVER

I'

COO/I 938-TV -9090-{)6.

2

Q ,I· a), Generate Huffman Code for the given image source. Calculate Entropy'ofthe ; sOlirce. average length of the code generated & the compression ratio achieved compared .to standard binary encoding, . [12] ,

b)

Q'-..'.

.

Consider an 8ilixelline of gray-scale data, { 12,12.13,13,IO,13,57,54}, which has been uni forml)' , quan\ized with 6-bit accuracy. Construct its 3-bit IGS code. ; . loa] \ a) Explain following methods of image segmentation by giving appropriate illustrations: ' ' " . I, [8] i) region growing (ii) splitting & merging.

b) Assume that the edge in the gray level image starts in the tirs~ column and l~nds in the last column. Find the cost of all possible edges,and sketch them. Find the edg.:: corresponding to l1l,illil1llll1l co~r path:

/

! t

[12) :,

11)

a) Write 8><8hnr!' mntrix sequence of length n=-8. Write walsh mntrix & compute

& show bUllerlly dingrnm to compute Hnrr coc,mcionts of a [5] wnlsh coefficients

, .1

of the following :2x2 image:

[5]

4 .. / I,

Compute d)

hadamard

Transform

Explain how Transform

for the sequen~e f=(

encoding

helps in achieving

1

023

I J

1

r.

[5]

Irl1age Compression. "

Q,7. Write short notes on any THREE offollowing':-i) Homomorphic Filtering ii) Connectivity of pixels jii) Discrete Cosine Transform iv) Hough Transform v) Compression using LZW method.

4

.1.""

[5]

..

, [201

".;

I

, I:(t "

\

/

i



I, I Pro I~:!

ion. 5080-06.

ec&:~

l

ru' )(

(REVISED

[j: MA6E LB. (1) . (2) (3)

'.: . I,

J! ' ;t"

;n,

i':)

.. .

l~:i.-

,

20

[

v

oj,' 6/-, 11 10 8); 9X 6. /7 /j12

X

by 4 oits/pixel

hns loilowing

8 8 4

gray level distribution.

12.

4a

16 17 U7 .11 X ;1-' y.; J: Gray I Levels

Modily the above Histogram .

such that the desired

rx-I i,/ 0 2

Jislrlbuli''':11 is d:; lollows-

-.

/7

,.

, 1

_.

, (b)

Draw Original specifiod and modified histograili. Why? Fourier transform and the irequency domain 100ls With tl10 help neat block diagram

':'.:

'., ".: . '~".' ",(I'

Explain.

Suppose that 8 x 8 Pixel P image represented

17

.•

~~~ alwee

(n)

:~ :i

: 100·,

(c; '

3,

.--

r·)Rcct5::Cl/'leD

Delino "Image" as a two-d,imensional array ancl 51ate v,n;;: I:. ,Il:;>;!' (lY Gray Lovel Distribution. Call 'vori<.lbie-Iengtll coding procedure be used to cornprc:;:.; Iri$!'.J\)'drn oQunlised image with 2° Gray levels 7 ,Explain. LoVi pass filter is a smoothing filter. Justify.

('..,)

!I

•••5845

Answer the following questions (any four) (a) Explain why Brightness Discriminations is poor al Lth" level 01 Illumination. c".J<€ (b) Explain convolution in the spatial and the frequency domain. Derive the relalions"i the two domain .. (c) Compare between Contract Strotching and Hislogralll Equaljzation. (d) Explain the role and, natura of image lidilily criteria in image compression. C;:"i" algorithm twice, (e) A 2-dimensional OFT can be obtained u:.;inU iJ 1-dilnensionai

,h:

..

c. '2. ()

Question No, 1 is compulsory. Attempt any four questions from the remaining six questions . (3 Hours)slate & ju:;tify IIw assumpli Assume suitable additional data if required.

j :' ~11.. ~I ,,;

D.I2 COURSE)

(
I

iJ.1~

explain the Bnsic of filterin9

Give the reasons of shifting the Hadamard origin, What is the difference between matrix orthogonal?

ill

the frequency

domain.

(b)

If the kernel of an image transform in matrix form. Explain.

(c)

Write Kernel matrix of size 4 (four) and application uf SI,w! and I Iail( Transform, i/nngtl f(x,y), find the Slant and .,Hanr Transform rilalm.

and symmetric, '

5

aJeQ.... ~nsform 7 Is ~0.n'IQ... the Hademardl

~ and Walsh~ Hndamurd ,

is separable

8

:;u u~uful for image enhancement.

the transform

can be expressed for the given

5

5

,

3

Ir ,

.

i

';>"~J .':

:5.··.

r (0)

, . ..... \:. .... \.....

f(x,y) =

lI

44

6

46

G

2 2

5

3 6

I I

['

" :.>

7

:!

57

1

o2 ]

Explain tlia LJu!llc principia:; (i)

I (c)

i

1_.

'I

,

Doloet tl10 boundary and segment of tl10 givLln illl~VLl us/nu ur,l~h Illoorutical tho grapl1 (or tl10 glvan Imago .

, 1~';~.::!

I

]'

2 ~

(ii)

01 dotocling

tl10 (ollowlll!-j III 1/11.1\101-

Point L1nos

I : (iii)

Edgos . (iv) Region Spliting Describo how Hough tansform

is used for boundary '""""

c'5

'

shape detecti

approach.

I\lso

draw

G

.. :;

\../~I'r

153'

Con.5000·YM·S845-06.

c.

(~)

Explain

(0)

Explain the basic principles of transform coding for irn:lVu C:Ul1lpl~ssion. "'ustr.Jte the help of OFT and OCT.. Construcl the IGS code of lite' given gray IOV01 f.l:l : Gray lovel·- 75,1.05,98,175,200, 1:30,210,220 Decode tho received IGS code 0111.

(c)

7.

Write short (a) (b) (cl (d) (e) (I)

the Origin 0(' data redundant.:y

in digitul irnayu$.

notes on any four of the following ;Median filtering;. Connectivity of Pixels. Zo'oming of a digital image. Uniform and non uniform sampling. Predictive coding. ' Fast Hadamard Transform.

! 10';" tll.!:;\.: I"v<.JuCldCJClci(;)s uru reduced? tllC

same with

20 Ii ".;. , j

.';.

l

','

'. y ! "

!:

Related Documents