0% found this document useful (0 votes)
84 views41 pagesDIP Unit-2
Uploaded by
karunya mittapallyCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
84 views41 pagesDIP Unit-2
Uploaded by
karunya mittapallyCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 41
=
—)
utenti (<‘éé‘ité‘ SC‘:
IMAGE ENHANCEMENT
Enhancement 1s a paccess in which we ane gelling
belley vesulls were as cemparecl to the oniginal trade
Tage Enhancement *~
Jr0ge lO ana
This %$ @ process “in which we arc gelling, beer
image tne weae 05 Cormpawed to the oviginal image.
classification of Image Enhancernent -
the image fnhancement i classified whe ‘em bypes.
1) spatial derrain Enhancement
1) Peg. dornain Enhancement
_ spatial filleaing-
whe spotial fillexingy elp 5 sepresented by geny) if dhe ip
<9 | image intensify. value (2) 15 den” Then ye cosnesterding
| ob image intensity» value (3) 5° baghe.
txt Ube tip srrage intensily value ‘x 3s batght (v-D-then,
‘the coxsesponding? ofp tmmage intensity value (50715 dosk.
These 4enage tnfosmation of «v) lyanstormation 75
used tn” Medical - field
#In osder te analyse x-Ray image of " pacast Cancer
tissue”.
$$ wey ae Bienes cee eee eee
‘ "
m4
[deal cox) identily Laanefowralion’-
| Ih An Lyansforrnation there 16 0 charge I the
rmage ds oP
| mlensily value of input -
| ae
i 3Yy
| | oh A
| ;
YN
| L
ott :
th Ye 3 Let
| ——
Tnpul ty?
ransformation cuwe Shoo
| combining, the ideal % -ve
iin above figure
| #2, ~ ansherrabion’
A leq, bxanstoymation can
| Fe cbycvn] > Conolant
be expressed as
where , $3 ofp image intensity value
Yip n 1 Rby Considering 7
| C=) 19 constant
||# By Considevings. cz! X #0 * We get the covvesponding, op
image indensilyy value -i-e> BY gubstitubing- the clef ferent
values of fp image Intensity ex ty? in the above
expression g& constant te’ value 19 4. Then jwe get
| dfesent of Tnbensityr values.
wonit-2y LAN
££ |
1 ip ienage tle
|
| ¥ Fore the Lt¢, we tn conclude thal the rano- ange of |
th image intensity values one mapped into widex - vange
cf blip image intensity, value.
# he. opposite functon of dsansfeymation log-cusve 15 called |
as " Invexse by fransfosmation cuave”. In this » wadex ange
cf ip wrage imlensity, values are mapped into navyocw
lk . \
yange of Gp tmage intensity value
|
3)powex - - law byansformation cuzve .~
EN ar
essed as,
# the power - law lyansformation cuzve Can be ¢xpa
= of “tmage indensilye value |
. |
whexe ,
| Ys tp
| 6 Gnstant
u n
lain dg Constant velue c=! & by substituliny the 4/p
image intensity value cn and ‘Kk’ values. we get ofp
sntensilys value ts!
| 12) = nth vot
nth pocwea’
| image
X71 ,
06
ix foy ¢& x21 , then »we get
| x= 0.04 i
| X=a.l0
X= 0-20
X= 067
al
Yebs
Yrs
x25
¥=lo
1295
|B Pom the figure swe can conclude that ‘x value 15
less than 4 [ver]. Lransformation curves jrdlicates nih soot
Plt sc, where ‘> values ave shown above |
1)
3Yy
[X se-04 , 0:10 » 0-20 040 » 0-64]
(®t these nth scot powex low taansformation in the nowace.:
| vange of vp image inlensilys values ate mapped into wide
| yange of ofp tmage imtensily, value.’
i* For 'y’ value is guectey than 4 [xl] Boy | swe get nth uo
| Power jaw fxansformation cuxves by, substituting, diffexent
| values of ‘for a pasticulay Wp tmage infensityr value
| ane likewise for different values ef tip image imlensily,
| values & for diflerent volues cf x are y=h5,2515,10%
/*By observing, that nth ook power q.¢ .we can conclude
|
that wider vange of itp ‘tmage intensily values ane
Mapped into raxvow vange of of image antensilys value.
onik-2) 6/4) |
HTL HS % M552
\ 0
11 ol
Vert ty? : \
y' value 48 equal Lo 4 fx21]. then swe get "Ideal" |
anstormalion cave fo diffeyent values of ‘lp image {
intensity, value.
|ece: wise Ineo tansfermation feoetons
|
yon 19 pieeee |
| ore of the simplest faansfowmalion funct
| wise fineay transfoxmation function-
|2)Conbsast gtyelching (change in background) --
wees
\
| ert js a paccess in which the contsast can be spread
tn entize xange of intensities
device -
| let us Consider kwo
byancfoxrnation Cuave-
TF y=$1 & ¥Q=52 then »we can get
curve. Due to this there 19 No change
then , the corsesponciny’ #
porte. (49S: & (aga) alony’ the
|ineay byarsfow’
mn output:
yansformatio!
curve a5 shown below.
—
My
of an image: $0 thet |
Spans in an enliye jmoge of vecoadiy cor) displaying?
mation| |
(484, ¥e550) p's inkensitd»
+ Tha order to ovevcome the indensily aréifacks we Can
Consider (+1) aaa 10) Ww (a$0) 2 wax stt)
“oni t=) ae 7 _—y
# By using» the abeve franstormedion auave on to the lex
Gnbrast image we Can get the sesulling 15 a5 4
binayy? tmage.
b)intensity> level slicing:
Ree aaah cas
¥ Indensily > level slremg, ell highlights certain portions of
an wage.
* for this intensity level shoing?. we Gan use two
themes of tyansmation — functions
# these indensity level skcingy 48 used for identifying
water shades in a salelile smage & fo find flows in
an X-ray mages .
# Fem the above faanstormation cuyve we Can conclude
that the ofp image intensityr coes not change fom cto
ONt- By Ae lo tt
+ the Of image le,
ae } inlensdy, value sudelenly increases whe
tip Image indensit i s¢§ when
holds sity» Value wseaches to A-pomls £
6 Same Ie,
same Irghe » indensilyy values upto Lhe b-Foinl
thot net »
hat’ yneans , in Yesalling, amage —bftw the a and B
Portion of intensity» value of s/
y fc of i/p
Ghicd at Hs op. sage Gan be bighk-
Ll
| /
| Mo | | 400)
° sr
‘ % iy >LI
|
# By absexving the paevious Lyansfoymation cuxve Wwe can i
Gonclude that the Wp image inlensily» wale oto A &
|p to t-l- |
* we can get the Isneay ofp [linear oly dhexe is no charge)|
# Beleween the points A xB the oulput trae intensity)
|
value can be highlighted.
[et plone seme |
# pels ave digital numbers composed of bils. Fox
example Lhe intensi
iby of cach ptvel ina 256 level gzay}
gcale image 15 Composed of ‘8° bils.
£ Instead of highlighting” intensity level xanges we could |
highlight the Contajbution made to total appeasance by |
bik ay Sh SPeeitic — byictge
Fan 5- bit wage may be Consideacd a5 being, Pee
of ‘Sone bed Planes , with Plane 4 Cordaing sng, the
fevest order bik of all prrels in the image and len,
‘8° 35 all highest cxcler bils.
lég
¥An "158" bit plane csnlains very lower detalls of the
fmage in order fo incacasing, planes fom 158 to Msp
we can ghsexve details of a image.
+ the 'msB’ bit plane contains higher details of an image
# this smesthing» spatial filless are used fer gettingy only.
Smooth the detail of an frmage- This Smeother details
of an irrage is known aS Blinedl image. je, @ Streething
Spatial filer is used fox makings a qyay> leve/ image
ints Blurted image .
*® Averaging» filler can be classified as two types .
Smreckhing, fineast
5) weighted Avexaging» Filley
} SPotial f;Mey
i) onwerghled on n
* Meclian fille y ©) ardey 5 ftatic (ron-lineax )
ONT gl oy |" Amrmang” ft
of ~
qhis averaging. fillers are a lypes. they ane
weighled averaging» Gllexs
Wun -weighled ” "
gq averaging’ fillex xesponsa can be xepsesented tm lasing,
way:
——,
ait 4 Ze
Ji a, Wi2! Pfoy 3x3 Mask]
| Smoothing. filleys ave used for blowing: and fos nots
yeduclion . Th ts Used in pye- psocessingy tasks »such as
yemoval of Small defatls fam an image (large) Paix
dgings of small gaps in hres
| fo object extraction , and bat
(oy) Fustves - ‘
fineay
|% Noise Yeclu
clion can ‘be accomplished by bling 4
flley ancl also by nan- linea filley
Linegn sreothingy spatiol Bile
filles av the examples of linear Smoothing,
— Average
| gpatial tillers re 5 of two types -
| weighted averaging? “Aller
| VY onweighted averaging fltex .
lek us Gonsidey 3x9 avenaginy fillex , then the
Micient volues of ag filter 1 98 shown below
Co-e
_ fig. Unwerghted Averaa' 02? filled
bME-2) \ uw Fig. weighted ave xooseny fille?
In umeighted avezoginy’” filer , we ave egual weight
age fo the co-efficient values of mask » Fxom the weighle
filler we ane Coneludingy fhak some co-efficient
avexaginy
values of mask aye given by weightage -
By using avesagng? filler » we can get the yesultant
blued = jrrage gt)
Bf weet cuss oyre
96) = 29 eb
2k wot
ssa bob
aticulor
where 1 WG-4) fs a o-efficient value ot fa
| position (4)
| wheae » the range of 418 Ob 9-00 NI
| the " ny $6 O12 vee Ne
Tn oxdey $0 incvease in dhe dimension of moe,
| from 3x3 to 35x35 by obseavinyy those two yesultant
| images, we axe gelling, only Hovimam bluaved details in
n of smothing, spatial filler
| Qse of Maximum dimensio
|G 35x35- | non-linear Joadlew slalic. filleas
|* ovdeystati 7
7 ic fillers ave fillexs whose espanse tx based
on ovdey of pixels
Contain in an image encompassed by
a filers then the viddle piel value con be reprexined
x ae pute value , when we ove using fillea Such
| éype filler 5 called a3 median oy orderstolic Giller
ix This oxcerstalic filler aie used for ¢leminaling salt
and peppery wroise in anys tage ive, this ovderstat ic
filley aye Suitable for elem inating, galt and pepper
| noise in Qn jmage.
[k lel. us. Consider 3xa Mask in dhis Mask -In_ the
Pixel Position is 's'. Consider piel values
[10120 +201 20,15, 20, 20 1 25,100]
‘paccording, fo the orderstatic filler . the widdle
value Can veplaced with newex value
by aaxanging, pive! values in ascending oxdex
=o |
To 15 20 20 % Bo 2 2 toa]
i* when we ae considers
ts. the middle pixel:
shenpersoge ood flea
“xshaypening, _spotial filler one used for getiny th
» delaile in ndensity valve of details Jn this »we
iation (next pirel -P
diffeaent
dffeyential o¥ different op dong y- dixection
yepresented by z.
Peete vvit-a 1]
middle
pryel
Ihat 1s obtained
ngy 5xs Mask 1 la pic) position
Sharpes
Gan U9
wesent pitel = one diménsierel
fan be
re My
|
(* ay = fOr) = fla) fast adler, clevivative
+e fern) 446-0 ~2f0) 34-D 34 order devivalie
je a- -D fissk over devivalive along? y-direction
> fly an -Fly>
2 flyer 4 Fly —af(y) wd onde
| . dexivative
¥ 2-D fixst oxder dexivalive (along x~ ditvection Neclizection) |
3 + =f Gey) ef Go ye - FOOY) ®
’ ;
ve, ve yy oF FO- yy) +4 doy
RA stespening spatial filleys ave based on dex
“1 4400 y4D + sats) 4 $0)
va tive
let- us Gnsidex a scan line alonz, x-disection.
lele |e |e Js ‘4 | ape rLefele fel e| 66) 6G,
| Above Scan line shows sntensity, values he “intens: silty
profile of abeve Scan line.
Gnst ==
|
der i
*In i jer lo get the first axder devivative for the
above Sean line we should Consider the follower Graltfions
grist oxcley devivalive ofp ts rex for crolant value of
gnlensily +
x Fast orcle’ devivalive ofp be won -zewo for on-sel
values of xamp and slep.
4 Musk be non -2ex0 at Lhe end of amp Ga)ster:
6 ih iii le ele 66
k Je|¢lels] of slab}
#d-
DP 00 © -I-t -1-1-l00000 50000
# The Conditions for getking- oxder dexivalive for given
San line.
“gyMust be zeso ot Constant intensity values.
#1) nf Non zero at on-set values of Yamp (ow) Step
om) mast be zeve at the end of yamp (ox) step.
=9 |
(ele fe fe [s[ al af al TT [Tek lei elels
16 090.5 75000
o ol ooo 8
x 2 fom-t&®
% :
of = foiy 4 fon)
y- fd)
dvace the qvaph of first and second orcley
x led us
phon below:
devivalive ofps a
omit 2 / ts] Ky | ‘1 exjvalive
“Teage shaanening» by wsiny seeel geley AALS
‘ A lesivalive of
lev
)
LY .
US BHeLP)
‘ me,
+s #! ouler devivalive
' FIs 24 oxdew devivalive
5 «
6
| Clapbeiard:—
ry of ot
vFs ae tay
of
ye = $04) af Gey) Ha day) tdireclion
i )
aye = POH YD AF COND ~ Af OOY) Ye diver!
Sovp o Ly oy
|
|
%
ax oy
|
®
efGeH yy p FOEN yy apy FE GE ya) ACY 1D 2¢Cea)
PFs Fey) EF OY) FPO YI AFG -—D “Mflny>
| :
# mow [ets us place the w-cfficients value of above
CXPYCSSION tn 3x3 Mas oY cindow :
ol o
4 foul)
oO t
9 I 5x3 i
onl L lol qt bove sk
* In above Mask swe Consideatngs anly ly’ netghbou x =
above Mask 15 alse dapwn following 04
ee ilo
fe | 4 '
¢ |0 “1 o|
# Gimilonly » by considering the diagonals we qet Mask
to-esticien!s values in the following, “a4
| t I ' |
L' | 8] 5
a ' roo
#]In above Mask , 7s also dsown the tve middle Co-effici-
énts — values. ;
a shay peningy ‘image can be obtained by adding
Sharpe rings details 40 the evigiral image.
= |
[pee = Hog) seven)
t3 based on (WE 0? -ve)
where , the ‘¢? yalue
| Mask
riddle co-efficrent value of Shaepening” spatia
nit -2) ALLY
~ eerrre ’
TUS? Maskings and Iegh Boel Henny
* hh dhs eis Masking» and high Boast filleaing,
without usings ghaapening, filley diveclly we yan cl
shaapened ;mage.
Sh order fo gellingr that shaspened trnage without
using, shaspening? filler pwe should use the falling,
Prccecke steps
Take Blunred tmage $0.4)
‘i)Substaack blusvedl image Flu) fom oviginal image
fimy) 2 19 . Fa,
Suase 0 =f (oy) -Fary) -
Kwhen ket, then obeve xesultanl tmage 15 Said fo be
on -shaap Masking» tmage.
*Tf kor, then the vesultant tage 9004) 15 Said be be
high boost Fieri, ymage.
* where $0.4) =oai ginal tmage
FON) = Blusved 0
Fos 04) = un -shaxp Masking, filles
using gh exder deantives fox tion -lnear) image
Shaxpening» (gradient operatons) .~
# A 4 order devivalive of image Can be yepaesented by.
oe B-[F]
x
wheae, vf = a oadex derivative op of an ‘mage.
x = gyadlené along x - direction
fy = gradient alongr y - cltvection
wnt - Ly is]qy ae
yNagnitude of 4! oxdexr dexivalive is wepresente
ordey
n
d by,
Moogs IFT = J9%49,* Ene
y
Jn oxdey Jo make the above Magnitude of ot
the above ¢#'n
| wale ib
as,
MOOD 2 [gal 419y)
devivative fo won -[meay
2 [3 +25)
wher » Igy! indicates Magnitude of qyadcent alony
x-divection
Con be
Igy) inclicales Magnitude of gradient along
4 - divection
# A gradient opexatoxs ave
+) Robext opezator
ii) poewitt opezator
1) Sobel operator
Gackent by using Rover
of
's) types.
A pobeat opexator Consisl
1s usecl fox getting gradient °
another Hask 16 used for _ getting 9%
pe] . Lets
T
| ?| ta 4 0 |
ow, lek us Corsider an image
wn texms
of 3x3 dimension
alony -X~
UXQ
best opeoles
of “kes Hask,-one Mask
dizection and |
adtent_ oly Yate
intensity, voles z,] 28 | Zq]
tar on to the image.
*noce, let us place the above mask
diyection
~ ¢
Now , let us place the above mask (b)on fo the tmage.
We will get qyadtent values along 9-divection.
dy = 297% ;
“Therefore , Magnitude ef qyadiené ts xepreserded by gx
MOog) © Igy) tly]
| Moy) & 2q-%1 4 [zg-%)
" 5 = (t-1) [ecards |
a whexe , w= Dummy. variable by using lejbriz's wule. we |
@n get the gol’n fox the above eg'n.
jLeibyizes mules
the desivative of diffexent integral with yespect 0 its
19 jnteqzal evaluoted at the limit , from the
| upper limst
basic caleulas, . 5-ry)
& od
ae > Gy
| = ang [fi peordw)
| LS $= = = (- PPG) i =
wrik-2y aay _
dion 4%
l “asibulion “hi
y .
| =o uniform probability
, dista;bution Sunction
Gntincously
lel us nsider a 3-bil image whose dimension i5 64xéyz
a Given 3-bit image , M=6u , Nzéy
Perey Mi,
[k= PORIXHW] > Gaxey = von,
ea nk POx)
° 7138 019
| ! to24 O85 |
2 860 O21
3 | 655 | ole
4 327 | 00g
oy 245 9-06
| P2200) 0.63
lo Sal 9-02
~onik-8) aq) 5k = TO) = ay & Pol)
$20:
ee
So = 4 Z Pray)
io
51 = 7 [Pyle )+ Psa)
= 7 (019+0-25)
$1 = 164y)
= 72 Fos)
J=0
Sy
21 [Py (1) + Pon) +Pr@dd
21 font 095 +021)
=1(0-65)
3
53 27 Pyl%)
ja
53
= 1 CPr(ae) + pytnd + Paez) 4fy(33))
= 76-19 0-25 4020 20A6I
unit a) aay).
ne ee ea = 3 CRG0) eR s HDI TOI) HOW
= Tong + 025+ oh rote +005]
= 710-89)
95 = 7 a FQ)
350
= 1 [Prevsd + pyloi) 4 preva) + Py(73)> Pay) + Fors)]
7019+ 0-25 40-81 40-16 40-09 40-06]
= 70-95)
é
Se = 7 E Py lo)
J=0
=T[Polmyepecnds — ~~~ Pras]
= 10-194 6Q5+ 0°) F6U6 40:03 40:06 40-03)
Se = 6.96
5}
uv
4
Tz Py)
jzo
=H [Pelo) e Pe Oi* ~~ = Fy cr]
A .
27 [0-19 +0-95+ O- WHOM + 0:09 FOGG 40-034 0-02]
nit hy Sal 4] Here , So 213351
Sy = 6-236
$1 =30853 55 = 6-657
o S2=U-55 5 56 = 6-86 37
532567 >6 5 = t SF
oie-21 34] ONIT-IL_4 paRT-B
IMAGE ENHANCEMENT IN FREQUENCY DOMAIN
Man ARN A ARR
i 1 DFT Drow > IOFT
L a L ‘
we. peecessor
|
/
$004) ay)
Gd) = Flv) Huw)
Where , F(yy - Fre9. domain yepyesentation of firy)
H(wv) = Filley txansfer function
* let us Gensidey a one dimensional filler dvansfer function
= Ieleal [ow pass filler
nono
2) BLPF —> Bublexwoxth
3) GlpF > Gruassian ="! " n
1) Ideal Joc» pass filtea -
she transfer fee fuenelion of an
by, 1 th DUN? €Do
= = |
Hav) = o tf Dtv?>Do |
|
|
whexe , Deuvd =distance fo the point from ommm
ideal LPF 15 ue presere.
Do = Racius of cixcle.
ane concluding that an ideal
distance,
and
Re Pem the above egn , we
the fweguencp G@mpone nts - when the
lpr allows
piv? is [ess than o¥ egual to Po CDlusv? ¢Do) otherwise,
| h doesn'é allows any frequency Components through
ore —_—_
* the pexceplaal view of Tipp daansfer fanclion and 2p
Mask is as shown im below
AHN) Hey)
| WA
5 vx
Dotuw? ie :
TIPF foansfex {'n peaceplual view of ne aera MEE
we to Diwy) IPF byansfea #'n
\
Dewy) = [cu Pare (v- ogy] a
the ola Poe of an image 15 sepaesented by
[r= é Plu)
U=0 vro
Disadlvantoges:-
RILPF is havingy one Main disadvantage that is yinging’
attfacts ane ocean when we ane using ILPF . we need
only blurted details. but by using of Fipe we ane get
Buaved details and vingingy axtifacds. Q
we Can veduce this vingingy anctifacts by incacasing vedius
| of ciacle ond also by using other type of fillers lke
| Buttereworth low poss fille and gaussian low pass fille.
ln Butte xeaosth low pass flex CBLPF).-
the bansker furclon of BUFF 38 wepucslentod by,
}
_
H [odv) /p da
|
| H(wv) =
Tonge 1 24] Dwhee » Dewy.
Do
n= oxdex of filler
jew the help this BUPF we can deduce the winging
artifacts by, inc¥easing, the gydex ‘mn! becaure weee the
do dhe pint
intensity, transition 15 Slow , whereas comporte to ILPF,
sntensity’ Lyansition ~
ay tron?
eos’ 2 Hat)
: Sy
W
a
vi
tuo dtmensicrad
+r o9e
“e Mt
quossian kes pass files [eutprl :-
(® the transfer function of a GiLPF
pean = eevee"
Percept
“8 ye puesented by,
=> :
wheal, ¢ = 7b indicates spseotings function of guassian
the pomnk of ovigin-
luv = distance $0
son te? ig sodius of Pa,
tb Qnsidesingy spreading funct
then the fansfey fn 6 woditied 95,
H(uw) = oO) | 200%
FRY qnczeasings the vadkus of Guacle upto cextain limit
, SE Gm yeduce te singmg oxtkificates which axe
| alenay with blusxed mage.
Occuss
Ont D | 3] / ow
u“
NY ope nage
Pewcep live “Hol
LPF: -
Haw = ' if DlWv) <2,
° if DCU >Do
J
Ht) = ————___.
I H [Dtav/eJ?”
I) .
| Hiawy = 62 WY RDQ*
|| Hpe:-
I
i) Hiv) = f' GF Dewy) >05
| © if Dtwv< p,
| Huy) = a a
I+ [D0 /paw]
Heuva = p62 er Roo
| function
Huw) = -umutev)
The laplacian filley is used for gefting~ edges of an
| mose that is higher details of an image, the transfer
of laplacian Siltes is %epresented by Hlww.
Where, Wv one the feq. domain vasiables by Considesin
@ wectangulay Centex of an image the above laplacian filter |
init 14]
%
tui?
1
— _ |
oD
Cx055 SECLION image |}
with an Order > |
Lhroveh 40 | | faansfer furclion is waile if as, 1
| , Lo ty 2. |
| Haw = -4n? [(u-Peyy (v- ef) ] |
i Haw) = -aND’cuv)
| wheae, Dew) = distance from — oagin
gethe Enhanced image 94) 15 yepresented 1 spacial dorain
5 as follows
Gary) = Flos) + C* FOOLY
wherxe, $004) = axigin image
V'Fo04) = laplacian +node
~ othe .. . yepaesented in Fu?
laplacian ‘rnage W4CLY) 7S
| domain ds ,
Weng) = PTS Hew FONDS
nm by |
|
Jee know that an Enhanced ‘*naqe gcuy) 18 ave
|
gouy) = foyo + c* v7 FO)
= f00y)-P*$O0y) Coc s-1 because how) 1% -¥e)
Finally , by applying jnvease discaete Fourier toansfor
» fecquency domain values we we get 1a spacial |
enhanced vou -
> -R(YF (av) 5
|
|
|
=a L
| domain
eer
i pew] Fu ws
gous =F! Pft-
|
| guy) = F! cau] Fu)
| |
mien) 1 S414)
|
|
|
|
Sure oy OTT AS ——
ny ad High Pregueney
Unshaxp Masking High boost fi
Emphasis fillexing .-
In dhis stead of using HPF'S dineclly we Car 0%
Hpr's in toms of LPF S
Lethe spacial domain vask 79 yepresentcd be
| Fmoste (up)? = I fap CP
| whee , fyp (oy) = TE sncicakes lower details of A image
d value 94) 19 obtained by-
|*A spacial domain enhance
addings a spacial domain ask 9003) fo the ovigin image
$0454) with a 'k’ Value. pee
9O4) = £009) + K* Sprache C4?
© S.qouy) = Ff pau 4 Cy- Hap (rv PO?
x the above expression defines urshayp maskin
if defines high boost Glkesrng when k7!
eT foe k* a HLpCuv) 7 F(a)
vit
y when ke)
and
g(~Hy)
"
M
gay) = FTF Ce k* Hp fan] rams
whatever the lexm is pyesent
freg. emphasis filter
* Tn the above expxession
in Suane bracket Hal indicates high
g4(uy) = Fi J okie KeHyp uw] reat
> @ntaols the offset values and cy!
ky
#In the above en.
|
Gntwls high freq. Values.
7 ~ Unik 38ly Homomomphic fillers-
Ne NR ee \
FA jfhammalion and yefleedemree
Product is ‘mibvadacedt in)
Tih oxees do implermyoving |
Simantaneously- intensity» remge — @mpnesion and Grtrool
tnhancement -
fey. domain procedure
* We Kroes Fhat an tage f0oy) = 4 ey) YOO)
SS FOOY) = F009) xouy)
where» Pony) < illumination component
YGOY) = yeflectance component
“3 ¥In ovdey to sepanating lamination Gmponent gacflector
Component we qo fox nataval logosithum
2004) 2A) Sees stay) SCO :
Pee he
eseed| J ace | S@y
fy) = joy) vey)
ZY) = In [Foy]
ease ei
fap
sin [iooyreny)]
20yy) = dni Goy) tin xO?
LF (20) BF Gow) + FC)
A DET 2oy) i sepaeserted by 20)
Zu) 2 Fi (un? + Fy (uv)
Suv) = zen) HWW
= [Rw thr (uv JH
3 sly) = [Rewv) Kw? + Fr(wv) Hv) d
=e ay
q of ahh
ae tsansform to
| By applying’ inverse powtio
piscocte
Ista we geb $00
-6C0y) = F'Goy) Fx"?
Trikially sed oxsgiral
aduaa] — fogexithm in order to sepayate |
‘Gmponent — and yeflectané @mponent |
K Gemilarly » SCY) 19 applied 40 exponential f'n ot the olp “
| exponential » we gk g004)-
1 4 |
Goey) 42°09) |
e {
gage $0 ath
we have allie!
s|lumination
2 90D) =
| all "
2 ets), XO
|
|
9009) = joy) YOON? |
J ‘ 1
where > jQny) = ello
% OY) = eX OY?
the Homemomphic filker’ hansfey #7
by Hea) = ky Xe 11-€° (vita r/o] + Xe
Ham) i e
eThe above figure shows
| cxoss - sectional view of
| hornomoaphic f
| Grains Components % &%p
iter which
BUND
kd ual uy Tends to altenuale
+ ferds fe
g where %