UNIT II IMAGE TRANSFORM 1. What is geometric transformation? Transformation is used to alter the co-ordinate description of image.

The basic geometric transformations are, 1. Image translation 2. Scaling 3. Image rotation 2. What is image translation and scaling? Image translation means reposition the image from one co-ordinate location to another along straight line path. Scaling is used to alter the size of the object or image (ie a co-ordinate s!stem is scaled b! a factor. 3. What is the need for transform? The need for transform is most of the signals or images are time domain signal (ie signals can be measured "ith a function of time. This representation is not al"a!s best. #or most image processing applications an!one of the mathematical transformation are applied to the signal or images to obtain further information from that signal. 4. Define the term Luminance? $uminance measured in lumens (lm , gi%es a measure of the amount of energ! an obser%er percei%er from a light source.

Page 1

A.RAJA, L/ECE, GNANAMANI COLLEGE OF TECHNOLOGY, NAMAKKAL-18.

5. What is Image Transform? &n image can be e'panded in terms of a discrete set of basis arra!s called basis images. These basis images can be generated b! unitar! matrices. &lternati%el!, a gi%en ( ' ( image can be %ie"ed as an (2'1 %ectors. &n image transform pro%ides a set of coordinates or basis %ectors for %ector space. 6. What are the applications of transform? 1. To reduce band "idth 2. To reduce redundanc! 3. To e'tract feature. . !i"e the #onditions for perfect transform? Transpose of matri' ) In%erse of a matri'. (i.e *rthoganalit!. $. What are the properties of unitar% transform? 1 +eterminant and the ,igen %alues of a unitar! matri' ha%e unit! magnitude 2 The entrop! of a random %ector is preser%ed under a unitar! Transformation 3 Since the entrop! is a measure of a%erage information, this means information is preser%ed under a unitar! transformation. &. Define 'ourier transform pair? The #ourier transform of f(' denoted b! #(u is defined b! ( ')u*+

,
.(

f)-* e ./20u- d- ................)1*

The in%erse #ourier transform of f(' is defined b! ( f)-*+ ,')u* e /20u- d- ....................)2* .( The e-uations (1 and (2 are .no"n as fourier transform pair. 11. Define 'ourier spectrum and spectral densit%? #ourier spectrum is defined as #(u ) /#(u / e j0(u 1here 2')u*2 + 32)u*4I2)u*
A.RAJA, L/ECE, GNANAMANI COLLEGE OF TECHNOLOGY, NAMAKKAL-18.

Page 2

5)u* + tan.1)I)u*63)u** Spectral densit! is defined b! p)u* + 2')u*22 p)u* + 32)u*4I2)u* 11. !i"e the relation for 1.D discrete 'ourier transform pair? The discrete #ourier transform is defined b! n-1 #(u ) 12( 3 f(' e 4j25u'2( ')6 The in%erse discrete #ourier transform is gi%en b! n-1 f(' ) 3 #(u e j25u'2( ')6 These e-uations are .no"n as discrete #ourier transform pair. 12. 7pecif% the properties of 2D 'ourier transform. The properties are, 1. Separabilit! 2. Translation 3. 7eriodicit! and conjugate s!mmetr! 8. 9otation :. +istributi%it! and scaling ;. &%erage %alue <. $aplacian =. >on%olution and correlation ?. Sampling 13. 8-plain separa9ilit% propert% in 2D 'ourier transform. The ad%antage of separable propert! is that #(u, % and f(', ! can be obtained b! successi%e application of 1+ #ourier transform or its in%erse. 14. :roperties of t;iddle factor.

Page 3

A.RAJA, L/ECE, GNANAMANI COLLEGE OF TECHNOLOGY, NAMAKKAL-18.

1. 7eriodicit! 1((@A( ) 1(@ 2. S!mmetr! 1((@A(22 ) -1(@ 15. !i"e the :roperties of one.dimensional D'T. 1. The +#T and unitar! +#T matrices are s!mmetric. 2. The e'tensions of the +#T and unitar! +#T of a se-uence and their in%erse transforms are periodic "ith period (. 3. The +#T or unitar! +#T of a real se-uence is conjugate s!mmetric about (22. 16. !i"e the :roperties of t;o.dimensional D'T 1. S!mmetric 2. 7eriodic e'tensions 3. Sampled #ourier transform 8. >onjugate s!mmetr!. 1 . What is meant 9% con"olution? The con%olution is the mathematical operation of 2 functions is defined b! f)-*<g)-* + f)(* .g)-. (* d( "here B is the dumm! %ariable. 1$. 7tate con"olution theorem for 1D If f(' has a #ourier transform #(u and g(' has a #ourier transform C(u then f(' Dg(' has a #ourier transform #(u .C(u . >on%olution in ' domain can be obtained b! ta.ing the in%erse #ourier transform of the product #(u .C(u . >on%olution in fre-uenc! domain reduces the multiplication in the ' domain ')-*.g)-* = ' )u* < ! )u* These 2 results are referred to the con%olution theorem. 1&. What is Wrap around error? The indi%idual periods of the con%olution "ill o%erlap and referred to as "rap around error. 21. !i"e the formula for correlation of 1D continuous function.

Page 4

A.RAJA, L/ECE, GNANAMANI COLLEGE OF TECHNOLOGY, NAMAKKAL-18.

The correlation of 2 continuous functions f(' and g(' is defined b! f )-* o g)-* + f<)(>* g)-4(* d(

21. What are the properties of ?aar transform. 1. Eaar transform is real and orthogonal. 2. Eaar transform is a %er! fast transform 3. Eaar transform has %er! poor energ! compaction for images 8. The basic %ectors of Eaar matri' se-uentiall! ordered. 22. What are the :roperties of 7lant transform? 1. Slant transform is real and orthogonal. 2. Slant transform is a fast transform 3. Slant transform has %er! good energ! compaction for images 8. The basic %ectors of Slant matri' are not se-uensel! ordered. 23. 7pecif% the properties of for;ard transformation @ernel? The for"ard transformation .ernel is said to be separable if g(', !, u, % g(', !, u, % ) g1(', u .g2(!, % The for"ard transformation .ernel is s!mmetric if g1 is functionall! e-ual to g2 g(', !, u, % ) g1(', u . g1(!,% 24. Define fast Walsh transform. The 1alsh transform is defined b! A.1 n.1 9i)-*.9n.1.i )u* ;)u* + 16A B f)-* 0 ).1* -+1 i+1 25. !i"e the relation for 1.D D#T. The 1-+ +>T is, (-1 >(u )B(u 3 f(' cosF((2'A1 u5 22(G I)6 (-1 In%erse f(' ) 3 B(u c(u cosF((2'A1 u5 22(G "here u)6,1,2,H.(-1

"here ')6,1,2,H(-1

Page 5

A.RAJA, L/ECE, GNANAMANI COLLEGE OF TECHNOLOGY, NAMAKKAL-18.

J)6

26. Define ?aar transform. The Eaar transform can be e'pressed in matri' form as, T+?'? 1here # ) ( I ( image matri' E ) ( I ( transformation matri' T ) resulting ( I ( transform.

2 . Write slant transform matri- 7A.

1here, a( )F3(2 2 8((2-1 G122 b( ) F(2-8 2 8((2-1 G122 for ( K 1

Page

A.RAJA, L/ECE, GNANAMANI COLLEGE OF TECHNOLOGY, NAMAKKAL-18.