Pa Ho

AssiGNMENI-02

delecton In dotail.
Explaim Maar-H1ldreh edge

AgMam-Hrldoudh Edge_deect

Mam Hrldaeth aquod that

latenaity thanges 0 Kat dependant fmag 2al

so thair dekcfion Htquiea _u o decdtna dipent

)A
Sae3

mghh
uddeu iuitud
fist dRinte
chaug uu e
atqinlnty Zen
nse to a ppak

Choing 1n
a
h oud dauvaliue

Ma & Hrldreth arqued hat tha moa 4atigacany2orata

he
Hhox toutions i tha-ilkr v
whar
Glingthae

Cauian unlc

lo nd
an turman yG u go -alloaing

bfeanhiahions
4y)= SGb)4$'Gla
S?
72r2
S-y
Sal2 Sy -2
2
(olleching

his
kms giwea

equahen
Ha

lpdazian
-ial xpsiss

ausiau
q)
 e Man-Hrldeth algerithm

loGftkuhan out imagu, fa.4)-

cosia oenuoluing he

g.y)v'giy)rftLy)
And thn inding Hhe zon coMings of gIPb
oeleamine he oedgea oy
locations

Indicahing hat u can shoath the Imaga ist uith4

ha
ilts & han tamput th Laploutan of

K
guassian
heslcha can beummaxised a -allao

Rler th nut imaqe uith an NXN Gaukian leno

pass erohtained by sausplúng

a. Coputa Hu laplaan a inage
3 Hndthe cmhing_otha maga 0

3Explaiu poiu iue aud adge delkection rhnique

Ans
Delechion ISoladed paints
foint dechion ispALd an fhu Aecond deauAalieha
laplarian ecand adar deuiudiue is wed

n2 ay2

whette axe obtadnecl wang t ollustg cqu

eau ()becomea
2fay)=£aLl,y)+t(H,y)tfl1H+1)1{14-D
4fia)

he avptikaan an h înaplenmentad aa makUaina

he laplaiau maik,w lay Haat aat laas besa
 PN.
Das

eheoed al lecaticnC1y)n whish thu maaka cenle

In da,
hollauing
aher
en the oput obtainsd urihg-

o-thasuise

Lome q is p_mag I4annagatiuetshold

Hho
and RAgineLa hy- REw2t
tigt-ta24
k=
lhemaulation imply MLauiehe uighked diffennsu
hlu a pixeli neighbos

Line Delection
The masks_ line delectok_ak shouun below

-LEI
hongoutal
2-1-|t45° vesfhcal
-H

-45
lef RRRR donole heieonse
a n tha mask a
ahanunakoue, upojOL maska indiuidua lly

mrough an imags-
fa cestaln paiat in ha imaga R: lRi fa all j#i,tht
uhculuaoiut &daid ok mae likoy aoaiakd with a
hadzpnial line

be may be inkscled in doktchinglínes in a specifie diaethn.

le uHha mesk aoiaded uith hat diaetion 6 threshald

MR ae h
in e inktkeld
dinechion da-anad
in deleching

by
all
agiuen masku
lines on _an Image
hun ha magk

toughout inage acthold tas absolkule yaluue o4

 Edqe Dehechon
-txdklechn edges h an ma u haue to u
ist and Aecond axder doniuati a
lne -fint deaivahve is poitive at the point f aniin
& aut othe iamp asut maue am alt fo agki
iulo

alon
lantp&
hapla
i8 Zen
it s constant

ueas o(onstant
p poink
qmy
fn tha
leuel

Model of an Modal oatanp edg

iden edge

fig Ca) tig (b)_

he second denvaive L0ine at he tanition

agacinledutn the doank side tha edgs, uqatiue
Hho tanition askoualed uutth ha ight id ha edgz

zem aloug amp înt & oa a tehutant apu lue

iny-lesl pmilu

ca
detail naaa the edge,shounq
onaie

Am le pofle he siunt
Secong
cond deriatue e tha pmhle. deovaine

drom these ohsehation aamaomitude athe

ist denivative caun he d to dodert he prcesenue q-
an edg
a
t a
SD an
oiut iAn imag
wwd n_dekami whethar an edg
Siga
pixel les in u dak or lipt ide f an tda
 Paa Ne:
DeE

A hat mant ytda liakingExplai dau lining

A AAia lodal

EA inking
peccAins

Due to hose,bcaks in heedges fu On uniheea

itlumination othur ejfuk that intnduos inkeniaty

discotinuhes

T emove these nensihy disconhnuila he dgu celechion-

ae-pllaud by nku
algmthma padura o
aaoblz
edpixcelata muaaiuad edaps
0
Local lrocessing
All points -hat ase áimlar alc to setofhedeninad
aikna ak lntad aing an cdgropixda hat shox
hose czileia
zhe huoo pinaipal exies used ablishing imilaihy
edg pixela

D Sh-enqth fha pshanse of the gaudient oeataed to

pu dge pxel_
aTha dizrchion of gadient vacta angla)
A st properth the gin alu ar of
hus_an edge pixel uith to-adinake oyin_a ueined
neighborhood af (a Sinilar in magnihuda fo fho
Ielat ()if Whoe E
E

,
a4y-y{71ya) DenDogáte tegh

he diacchian_Cangla) af tho gaadieat uedta is aueaby-

dly)=tai9.
An tdgc pixel at
(a4) hau an angk kimilas
y)in
hepedehinadAaizhberhl
Ha pixelat n
<A
JAheus A ia
y o

alay)=dlMoy nounagaiue augu Rishold

 the-plaaing dep
nuoka
atal pacang kangl azy
Compuk H qadintmaqrluda
input Imag
M)&d(y)
mag
uhevalutatany pair

2fmahinany
tneinatesGy-

g) MCLp)2TmAl,g) AsT
otharuise

kue Tm i aHacthold. Aa kafsad ongla diaehon

Th definaa mnd oatuaptaba dechions
a 1
3Scanthe nuso4kfILCset fo 1) all gapeSets
nct xeed a secisied
kngthk
n ach ma hat ch
utale4h
dktet gaps in_ang ather direchion 6,
the hmzontol Kanning pmaduns in de
angikapely
Kotate the Eult baik -9

Lhat ke tho tpee nbizcuea canny dg dalectra2

Explaiu cann dl dakction algithm

fng
Ihe anny dae delect
Cannuls appmath i aud on -te hasic okkerfhucr

2dae paiuta Jauld be ull localized

3 ingle_edge _sois _esbonse

|lst Sayn denote Ana inpt imag G() danele te

Guussianfunction G0)=t
A Amocthed Smage an be anab taaualauing 1g)
h
This oheration is alauwed by Copuching tha adtat
maqnitude & diechion uing Mx)=G;+-
 PcNe7
lefm non mauma uppaMion
Specib umper of dicea mentation.
CLAL G Tun poabla ontations da
heyontol edg na 3X3 neigrhouhooh
nen maxima Auptesion
sleps
AAnd the de that
dineclion dsetto dlLu)
) Hhe alue of aaCy)h les han at afeat ona of ihà
a taighhorsdlong de,let 9a4DE0 suppucmion)

nOu Naima kuporeyed maqg

The final eation i h hreahald ga)to kduua

ase edqe_paiuks
he Cann edge delectm
Saamarisingq,

Smoath
a
he foput Image utha utian
Caapua the gradiout maqmikuds anale naga
e
3 Applynan mavima aubeeion to theamdiont

4 e
nagnitud.
caube
image
tbaeiholding &connechleky- analysisfn
deect 8 k edgi

Eplain hou ophimum oloba threxholding_can be aheud

byCHsus mcthod

Qtiraum Glabal Tn-esholdingUoingDtsue Method

Banciple mosinizing h bluo clam vaaianu
lt o,2---L- denoke tha L dishinct inkenaih leuds
in a diätal imaga kiza MXNpiela and let n
dnot ha numbeof pxel with

,
înkensihy_i
PnMN Cund
k=-hHshold valua, 4o Kti, L-L
 1-1
G _and C(k) 2 1-,(k)

he meaninensity yalu of the piau ainad tod

Jau
G m(k) 0 P(/C)= (k) Pao
hemean inlensihy, valun of he pixelu asngrad to day
m,()= teCt/ca)=

L,mt m4 Glokal maan vaua)

Betusen_cdas vananta A as
=P-mg) +L,(g-m4)
2 depihad

he optmun thtsn.old i Hhe volua, k hat

masimizus (K*), )=0KL-()_
nay

fAy)Ek*
Sebanhilimiawe

Ofsu's Alg.o
) Campude th ammmaliaad istg oha ioputimg-
Denok tha companents of

,1=0,,-,
2) empude the
.he histagraro
uunulative kums
L LRfa K=o,
3omputk ha umulaive moana,
4)Compuk ho alabal
Compuk
nlk)yor k=o,1,-ll.

ha b&turon-das
iohenaihy maan
vaxiane
4fo k=0,1,-=LL
) 0btaiu the otsuk threahold k
) Dbtain sehamblit meaxae
 hth uk of
cdgeininruny
Pa Ho
De

e
Edg toImprove GlohalTheahddag

an edge Image q ither th aqnitude of

Campule
theqradient, absduk valut the lallaian of
flay)-
Shecely a threshold valu T
Thmeshaldhe nageandpradu a binau4 Ínng
whichis ed axa maak fmagand akd
plxels mn Hy) nespaudiug-p "skarg
edgepixela
L a
mpule histogzaminq nly hechoien plzdh
in f
U he hisoaramom sep 4o segrmenat

fo-4 glakaliu

ing to inpme Glohal Thesheldiug

edaca
ne apDAch-tr irning he hape histopram i to
Consideaw thoz pixela hat lia an_a nA th
edges u_dba and tha baukqruund
lngmuunent is hat he histogiura uould lhs dapenda
an the relatiue ausooheub and backqmund
Cpaaks ioauld b san siu)
Wing thegradient _Vnlun u canfind edgs,alsa uith-
Laplauiam valu bondaning pirels on ohjet and bauk--
qmAundisles andekbminad
hExplain image enharkement Using Arithmehe Logi

Ofesetiona

Ans
Airthenatic oharnthions

our ut uetie opatioustHauy ana

day)=fuy)=pn-y-

i pddition_a ImaqAerngang
mage Aerngnq Can bu used to imace noise in_
a 0auen image
anidiranaisy fnagu al.y) med hy the addithisn
noise h(a4taan ooina mags {4le-
gy)=fCy)+ha
mparant applicahion ofîmag addithun in he
-feildofazholaq-
) Imaqs uacion
=An appliation otmaqa suhneton is în the
ment
nedical
a dfkenca fw imuge andanather
enhania

aea A
mdica fn.aginq callud mak made
-Given 9 imagei )and
ndiography
h{z.y),the dhferenta
blu tuwo imaqeiA pes as
gCzy)=Ay)-hlay)
ohfaiuad h n diferenceu blo all pais o
cOesbcnding pixela

Multplicathianad divikio
An Impatant applitation e mag mtiplicatina
C&duiian) is shadina Cometion.
(ansidur imugin aias mues lmaqu ha an be
moadiiadah pnduut4afait imag danekd
 DATE 7
by fxy,times a ahading-punehiona hty)
ie goy)faP1h{z.8p
-1f houy i noun obtain
-fauy)by ulbiplypng
the sensatd Imu y thehveise ahny)
ogica Opations
Logic aparations auralso_operatad on
a
pixel by pixd haiu

Jshendealinq ithbinasy imaq, ua ton xidarortorundd

-valued)4 backqsund (o-valuud) set opixel

)Nor logicaloheratoy

The Nor oheratioooa Aet A u the set of lenment

not in A
This tuns all elkments
opertion n A to oblack
all theelements not fn A to| Cwhite)

AND and oR opndons

-onsider à kgions(etn) A Bcomposed o
egundpixel
TheoR of sets is set of elemants belaging-

athey to A_ B n hath
-The AND operationa i host ademanks that

ak Common toA andB

AND

B
 (
and high bast leri
filkring
SExphta ushasp nasking
Ans
A proces Lsed fa many eatAin
h publishang

btrauhing a
oubtrauing
a

Snduusyto shaapen Imayeconisis

bunedEakicn an lImaaan he
Tnis anau,called shaap maaking
tonage
ed Hselk
as
4sty)=Pay)-f(ay
he Imaq cbtainaud
uwhee fx,D danata haspencd
nshaaup naikuyg
yia a uerson of
blumed fy
he izin ounshasp maskinq in daak poon
togehera
phutb

nphy, ohese _consiteclambin

poetiuelo And
blure ngatiue toa Comespandung
thn douelobing this combination to sazdssu shasea a
Imagr
A slitAuatkas gh&raliatano unsau makina
valiad high-haatHlkring. A hanst eied image
fhb idefinad at auy poink a4) as
hae
hs
AzL,as bta
kauahion mau be witön
a
blumed ueagian of

m equ", e abtnin

as heekaon cepbutiug ahiçh beost-Seed

imageL
equ (4)À pplitabla in genstal anddlo nnt stale

evpliccty hou th shamp imaqg i bttined

LL eleut to use the llaciän,hen uu now that-
 D PhaNeT
Tn this ast cqn LA)beones
APay)vfuyo conler
h (oeffiqent of
mak negahe
thb A ConLopaciop
(oeficiqrt of h
fHt{OLY)"Lopla.cion ailposdtide

1oExplain egionansed seqmentation

Aa
Kegion baied áagmentation

ahmach of amilanty ed.ed Aqmestation

The
ontionln hi atmach, kamantationi ae kased on
Andinghe teqions diaect
Tuo Region bad seopmuctaion pmacha inaludos
Reppan qrmuaing

Kegion Golithing & Merging

i)Kegion
Repon
pixels
hkeing-
Gtawinq i a pocedue
_ubtgona ito a lag
which _gups
egians based
a on Sohna
patcdefinedCHlenn

he basic appmmach to stast_ith asekof sed

i
poins and _ th geu kpions b akbending to uh
seod hok nughnng_pxels -that_hau kinilda

pmbeties to -tho selL

Seleedhlhen apro iofamnation à ant aualable Ae
pma dua o compa at_ALApixel, Hai same
cel o pmperties Hat utirnatiu iuill heuxd _ain a
pixela to egion duning h qmwing PR
4 tha kak o thox computatiank aheus cuktg
ualues ha phxels shor mpehes plue ham neaa
a Cenhmid ofHae& clusks on be used as seds

Sa saleehing-a se opmlen -ele.tionof simlaail

Oikna not onludopends sn nahuxa o blen_buit alsa
On tha ty oimagk dat avaulable .
 Saqntanta.tionoan X-Ray Jnag ofa uld
CHha hniaatru daak kaina) to dalcemine he_tak,

Ragien Spliting& Maging

hls aDach kubdáides an îmage nto t of adbi
disjainled iegian 4hen maz Aplakha ktpion

inadar fo katisy ku tquizedtanditionsf

Imag qmurcddtion

lhe splifîng Mernaing_can be uummangedhyalh

pmcedue
SHlt Intbbur disiaint quadrans any ktgen R,
which P(R)=£ALSE

Menge any adjiacont_eons RjandR fa which Plcu,

TRUE, (uwhen no fatfhar splithag iu poaala)
3.Shp uhan no tudlher meming-plihng fa3hla
RRa
a partionon@d imag
(b)Quad free g a)_
- evaMple One poabiliki fo split ha imagk

iulhally iulo a Att blocka kurthur splitingi

ned aut a dosChibed at maginq iu
ptuiauy
Juial to goupA ofou blocka that a
descendak n the quadtee kmeentatian hat saity

Hha peidicalap
Ihe mincpal
ituseas
Bntl ainal mtgig
aoluantng

Sama quadim o
Hais aproo.aeh
splithing
ds tha
4nagang
 P Ne35
DATE
Explain maga
nden deuultiueeobancamantuing
uking st
si ondar
oxder and
and sund
send
Ara
Use o Rst Daivatives -for
fastdeaatiuea Enhacement
fn qradiant
Gradiant
hemaantuda o
imagpmDceaing Cc
Impkrmaked wina
Impkrnsded uing
hogradiant
He grudiont at faa-unctionf4)
8D olumo vectm comdinala (4) dainad a
LGy
Ihe magmhadsof
his vectn is
qiven hy

(A)_
f iscommon prachce to
appz0ximale tha magoihuosof th
qpaoient by usinq abauto
values

2Other tuo dasinitios phopokd by Robedts in-hatanly

|deuelopomant of diita iaaqepmceanq
wse croA
dlferncoa-
q-lz4-15)and. G4=(28-2)
l4)-
hen gaadient a z=[(2) H4-z) |-
we ux ahsoluk_valua. v
|z-2,tz-2 9
her knaska ak kffeed fo aa Kobekts Cou
gadient
oentos Maska Len aeae auokuard o implamant
An appaDaimaion ang absalulo vallu,sihllat paint 2-
t ing 3 nki
itzt2)(z+zti
 seof Secand Daivafve fr Enhansarnent laplacian

ppmah baicallydapiningadisoeie
Lonsita of

nlafion th stcond adordaiv.athiu ond han

constnuhing ler mask lnied on tha mulahion
can
erot
be ahoan that
Ma
e
Laplacian,
aimplest Ikkotrfic

uick,fr afunttion
darivatiua

)
o h uanaka, dephada

lhelaplocian a linsar oberator hLa daivahu

A arder am lnuanoherattous
batial
fatiol second-derdeudvahiue in ta -dihtchion
=tLy)t fa-Ly)-2GLy)=
I3iny-dhchion,as
e digihial implementaion o aLodlacian i
obtained umkning a huo coaponen-
lqu

4f(a4-
Betau Hu Laplattan ia deivaiue oherat k uik
highlightsqau-la discontinuhu în an imaq and
deemphahze
Leuala
cons wih slau am
Tha baaic oain whihLuL h Laalanhn f
Image enhancementu a dlou
tf tha Cenler, coencent

Mask is vkgastiue
 Pa
Dare

Defie the follousinq in datail

Multiple thveaholda
Ang
Nulthikuel eshdding d pma thoteginats a gay
wl inag uto scuea distfoctInagca

his lechniqua dalemnea mant than onathreshdd tr tha

qivenimagand qments the Jnag 1tb cetain bigtrast

uural
kaions, uwhlch tomapond to Onabckgoud and

obiezla
n the asea okcdases, GzC2. g,ha behuaan

class uasuana i:2(ME-MG)

Luht H2 and m S Ih
he tiar thrthold Maluas kk k that
maiize (kk ma 0ksH lku,k3sškP-

fr OIK
thet clae
eharalad
onsiting oHhrce inkensaty inknauls

b hua neshalds) da blun claxvamana iu

(ahih

L(m-m4+(ma)t(ma-ma)

>The oghmum thraholds,

Theaholdl Imaqs i giuen by

uhee Qh ajeany tme vald inkensy Values

one

wh - haeshdd

total
elends drechy fomultiplk thareahdel

image vawion
 i iade aeahold
Ans-
Vaiadetheaholding Calsa adapie thraolding, a
which he Aalu
theshold vaniea ova aImag u
a tunthion of locaimaq chuacknshi, Can pod
fhe &oduhion in fhese case

echniques fa choosinq eiablu Hbxahold

Vanous
imaoePhafioning
Sulodluide an imaq into nop Ouelappinq htctangla
Ihe_fectangles anchoosen gmal_en.cugh So hat ho
2llumin.ation o eah asproxinauelay unifra
basedon local mage RocoMmng
let ay_and y danole te Standard dauiation_and

DRan Valu o fha set of pisels contaíned na

neighborhaod Sny,entered ot condinat3 y)io an
image helocal heeholdk,yathnzY
backgpound isneasl canstat ly=aGytbm4-
=uheR abat non negadiue conatacs
MGia global Tmagt maan
-The eqmaubd imag à compuitd as

im Usinq
lo
Mouinq Averngea
ffcH)ET
-Ihiesholding kauedan mouinq aetg unka uuall
uAen au

;
Hha obiech Amaluith apeet to the
Imaq Mze
Quile useful in document proceang
-let Z denoe inlenity of point entaunkred in
Sanning kequonLg_at Step BAIhe moving avetage
at Huspaint i gn h
m(k)=Ln =m[)+L2pe-Z
n
f=Kt2n
n-number of pojntt umd in Comchng ha uerage
MC1) Z