ASSIG@NMENT-2. p oO NAME: STDDARTH AGARWAL RTU Rell No! DOESKCSBUS. SUBTECT: pTP SECTION: €cCSE (GL) PART-A @.] Why we need +o Wompreds images in digital image Processing : Aus: Traage Commprersion is useytl be cause it decreoses He ArMount | MeMMo¥y Fequired +O Store iMoges digitally or Coramuni Cate there innroges over a networs Such as the intemet - THIS Wncesned with minimizing tre number dpwits Teguived +0 represent an irmage. Trnroge compression may be lossy 0¥ \ossless . @.2 What iS xegion- based gogmentation techniques > Aus: This process involves dividing the imoge into Srnallew Segments that have a cextain Set hmttes. : This technique employs an algonitiom trot divides tne image into Several components with Cornrnon Pi xef CharacteriSHes - The process looks out for chunks 5% Segments within the \mage - These ave 2 Hypes:- i) RegiOn-Growing TY Region -Spiitting . @-3 Explain hormomospnic Hlterng moctet . : AVG. MIS A QenesHized yeohnique for Signal and image. PrOcessing Hevawing Q@ non-linear mapping 40 oc Aifperent in Which Unear Filter tecnniques Ae applied, domain (domain, fatlowed by mapPing roaak to the ow gina + Hornomosphic filter 1S Sometimes Used For image enhancement ot Simultaneously Normalize the bxghthess ACeDSS an image And FNereases OMYASE - ~ USed 40 vernove multiplicatve nolse that hos cevrain Characteristics.BY Exptain image ssimation technique Using iMnvexse Litering : Trage Restoration — Ty seberS HOO Class a] M eines Anat alr AE seclucing oF yemoving various types AistoOsions ApMHeiMege & interest . Tmage fj Hevieg ~ sb is adetewminislic and aisete method for image sertoxation . > The Tmages involved must be lexicagag prically orelerect - ANA Means Tack AN image iS tonverted +0 a coluenn vector by pasting the wws one by One| After convewing ther to columns. Aus: Qs Wheat do you mean by Adaptive Median Fillevi ng 2 AW: The adaptive median filter is clesigned 40 elirminote the craweatks faced by the Standard median. The Main osivantage adaptive median filtex 1s the size EGthe kemet sumounding re Comupted iMege is vanable alte to Which bettew 0/P most iS Ob tained. @-6 An image ts 1OMB before compression ANd SMB Hter Compression. Find outthe tornpaession gato and saving percentoge 2 AVE! Given. Bofpse compression sice % image After Compression Size A trrege = 5 mes = IomB COMpreAsiNg wes = Before /ormpress1O0 Pitter Compression = 19 =2MB means 221 507). Saving eee compression dh @-* wWratisthe need % image Segmematans ? & xplai oO edge point and line detection process - Aus: Troge Segmentation IS the process dy partitioning A agi tat inneages Into Varsor Parts oF Tegions veducing the corn plexities og und eystanding the imeg esto machines - muolves Separating He foreszound Armithe EET imoge. Tt cowd alsoPe ee ——————— ee eoV—nmsm ba tkayDund Oy ALSO Mpls, b& Pixels Pased On VaNOus 2 Limnilawiires in the color or Shape. -% Disetas the Tegion bared image Segmeniation tn detail, Aus: Trvs technique employes an algowihrn Cornponent tat Aivides the image IM Several Components with Common PIX@l Charactrewistiog The algorithm can pickup the Jey level for summunding Pixels _ They ave 6, two types: ~ i) Region Growing = THIS Method Weurysively grows Segments byincluding neighbouring Pixels with Simflar Characters istics . ti) Region Splitting - The whole image is consideed @ Single magion, Now to clivide the begion inty Scamments 1 checics for pixels included inthe initial sgjion cif the followthe Predefined Set dy criteria . @-9 Write the difference between arithmetic Loading And Huttraan coding > AUS Axithmetic Coding Ruftman Coding + SFIS NOt & SHAHSHCat method. [SFIS A Shatistical method = St Yields an Optra) sesult - + dt Aoesnot Yields AN a eee. a OpHimal sesult - ° The ss cane oa * There Isone {0 0ne UWrrespondance between SOULE | Syrnbol and ad [correspondance betacen Source ee Syrateol Anal Cole Worl. @-lo Explain Impulse Noise (Saltand pepe). Aus! Salt and pepper noise ,also KNOWN aus impulse noise, iS a Porm 6 noise somehHmes Seen an aligital Images ed by Sharp are Suciden diStrtoness This noise can be COUL Inthe Image Signal. 2t Presents itself os Sparsely occuring white are black pixels.PART-B fe @ Q+1 What is image compression 2 Explain aoout the wedundancies in a aigitat image. Aud: Tenage Lompression - Image ompeession is an Applica hon Eh clara Compression trot encodes the onginal Imege With feo bits. SIS Ye process 86 TOD cE AG Ane Size m adigital image While wrelaining aa much 4 The goat S&irrage compression isto make i Reanemit ,and aisplay irnoges twithout Sa irnagets visual quality. Redundangy ina Algitat Image - Reclunclaney webers 10 "Sharing enra information bo BEpmopent A QUBNTY OG information So Hat 1S tne wedundanuy of date now aprly Mnis concept oNdigital images wre know that computer Shove Pre ienracg es IV pixel values So duplicate pixel values oF maybe rf Somekienes ipnage Nad we yermove gome &§ ne pixel values they dontt affect te information 4 a0 actialicnage . Data Redundency is one arne fundamental component && dda WM pression. TT Wees 64 Redundany — 1) Spatal Redundanty- Inspotal redundanay there is % P Pi Y covrelation between the Neighbouring pixel values - i) Specteat Redundany — In Spectrect reduncanty, here |SQ Correlation berwoeen aifferent color planes ov Uatity a possible- t eaiser to Sloze, cvificing the Spectaett bands. iii) Termporet Reduncdanyy — In jomporert TeduNndancy, fer. {Sa cowelation berwecn adjacent fram es Ih the sequence 6% images. @.9. Consider tre ‘Ranstnission sf Messoge "WENT!, tomprsing A Shing of Characters Er: a.3,Ni0.3,Ti O2,Wr0-d . Use awirmeric Coding to Wrmpress the Me SSage 9EN | WI COUT" ©) AU Caen! “Dr CUVAMMESSON af, ME SS cege Goa 5 Nore |) oi Ft on od 083 ont OBIG9 0-89 oar + ou 9. Bibor w O88 orea on bbz 0. 81584 : a 0-86 ona OBEY Lo. gisua n 0-834. n \ fe) —~___ on 0641940 eget ane ? eee one o8 0.805 3. BI4Ny for w A= Uppes bound —lowes bound = 0.9-0.8 =O.1 Range a6 Symbol = lowe limit : lower limit +t (pavb. 66 Syrnol ) Range 64 'e" = 0-8: [0-8-+0.1(0.3)] =0-.840.03 =0-83 Range Q'n' = 0.83: [0-83-+0-1(0-3)] = 0-83-0-03 = 0-86 Range Ait’ = 0.86: (0-86 40100-1)) = O.864+0.02 =0:88 Range ag'w! = O88: [ors terICo1)] = 0.86 to-89 = 0-89 Same fos E,Nand T. 3 Hence ane amtnmetic Cocke woads - (message Decoding) 0-81602 < Codeword < 0-8162 Tag = (Upper limit 06 tocte.ooael +t Lowey Vinnit 86 tacleword ) / 2. 02.)/2 = O-8I6I. 9 0.83 OSB, 0.8162 Tag = (0.8162 10.816 ° 83 1 o9 89+ 0.624-+ 0. gia + oe: 81602 IW w iy w o8 0.88 0-824 7 o.si6z + 0+ 81S84 t * | 0-6 0-86 OBle o-SIMY re Bisus n n fp n | o3 + 0-83 oe ogi} 0.8194 T e Cc | 0+ 809 o-BINyLO ———— eee AUS EXPlUH Image AEMOAHAH Fehoique ULEAD MYves se (6) {iering, 7 AWS: & Trverse (Wemng IS 4 Aclowrmashic and cect method (ox Image SeAMOI.CATON Volvedt Must be lone evs conveste. opApPhically Oackeaed. That AIO eolurng VOtror hy Jo calurans. + “Dre mages in THSANS Prec any LORNA Parting Whe sows one by OH aficr 266X256 1S Conve Ned 4 CONVveETNAg 4norm * Animmage oo Slee Oc IUD VECO SESIZTE BSSIEXT Te degvadalion pred eh 1S Oy whe amiayes crac and Ihe ac gacdadion Pe Huge bub sparse maiarlarX + 3 EWE in deat Qa Hea n\ Wall? = [lg - HEU Ven IN a ematis x form, Where neess 1S & YeEnms condition scaly Happens * Tre first derivetic a The minimization FUNHA| MUSE be Set to Te TO. VIg-HEN® = C9-HATIQ-HE) = Catster) uA) = QT-aThE-f THT +4 THTME = -2Hg + QHTHF=0 DUTHF = Hg, HTHf = HTg £= (HTH)T WT9 a DE HIS a Square matnx and i inverse exists then $=H9. @-Y¥ Descmive in bw ef various Lossy compression TechNiqu es . OTIS MeEbhodl sf clata COM pression im imrelevant clatter 9 AUS: Lossy Com pression - Which the Size 4 file is veduice by the file. Lossy compression wants +o less zelavent inforrmah™ - “There are 2 Hypes Steps process — ident fy unnecessary andmR TSN Ns Sa 1 CS I Ne SY ® STEVIE Quamizahen- Sh 1s the process dy mapping ifr Farr a lange Se& fo Numenecal O/P Values IN a Smalles set STEPPE? san skim Coding - Bb is The process of Creating a quantized group Of blodies 6f Conseutive Samples feorg Source 1/P and conversing it into vectors - The goat 4 anspor Coding \Sto decompose OF bean sfoun tne i]P Something earses to handle. Several Technique 6 lossy Compression Technique — + Chrana Sub Sampling —> Coloe yelated > Colow sedurchon—s cOlO® intensity reduce * fractal Compression —» Reassemble cdlor, elc. + Transher loading —> Average color . vectos Quantizahon—> Small Set 4 data. Q-S Deserve the imoge Segmentahon based on Hresholding. Al8o explain global threshoiding. : Fei Aus: Thresholding 1S one 6% te Segmentation Fechoiques Ma! Senerares a bINATY image fev a gray Scale IMEgJe by Separating Te jNtD And region bared 0 @ threshdld value Hence pixels having values greater +a je Said thzeshold call be Wwented GL rite OF t [Me OFF rege Godt Poe pmers will be blank OF O- ry Prreshold —> T , Segmented image a Oy). histogzarn So an imege =fluy). THEM iy) = ip tCuvd> Tand Jui) so: FCHUIS T olp segmented image hor only 2 Chasses at pixels —> = one having a value *% 2and > others having a value Sh O- Te basic global thresholdi agalgon thm Threshalaing— y al ieee finas Phe post threshold value So Seqmentive, i) Select on initial catimnate 6% tre Hnzesheld T- Wi) Segment the imoge using T io foe Jwo gxDups AS and G2y ASB consists oh all pixels wolHo intensity values >Tand QQ consists & AM Pixels with intensity values ST. ———it) Comptlte the Average intensity values myand ma. IV) Compute He Ned value dgtHserhold TOs T= (miteme] 2 - ¥) Repeat step 2 thaugh 4 UOH! the diff erence inthe Subsequent Value 66 T 1S smalles Hoan a pre-eletined value 8. vi) Segment the tmage OS gouy)= iif FOnv)>T and guy =o iPFOUWS T PART-C Q-] Explain \raage yeatoration and degradation model indetail. AVL: Troge restoration isthe process Of yecovering an Tage that been degraded by same enowledge 06 degradation finch iband the addaptve noise tenn. Thus in restration , degradation {$ modeled cand its Inve se Process {S Applied +0 wecove¥ the Original immoge. Degradation | $0ny) g On Restovati fi Cony, ration 4 function CH) = gcmy) Filter Fong) > Degradation TOW) Rextoretion The objective 66 image wesinration IS to obFAIN AN estimate O4 the Originae imeye FLLy). Here 1 by Same Rovledge 6% and Cry) ourpLe Bf Image fly) + Temninogy - # glusv) = Aegzaded image ¥g Gog) = t/P ov OP Imege wp DOny) = recovered oF el tered image wk WOuy) = additive noise texm Ln Spattak domain : g Cony) = HOM yID Fry) +m Cuy) In frequenty domain : 7 G (yw) = ACH) HUN Cure ) for vestoration Fyiu) SA cuiu)2 Determafbe Vhe Heufponan code a£eignene mt @ Tarcecilae: hor Ihe follevoing data . Sy mo) A B Cc D € i e Prvb. O'} Joy |0-06 | Oo! Jo.os | 0-3 STEPD. oy! o30 70.27 10.1), os) eee i Average length ! Lavy = 2 Lng). Pelte) k=0 =) Koy +F2KO-3 TY XT FY XO AYKO-O6 +4 XOOS = OY+O-G $O-4+0°4 FO-24 0°20 = 9.2y bits /Symbol. By Entopy He) = & - Pelean(se ) oy log | +o. 31052. 0-1 1092-1 +4 0° ea 0-04 o-l oe +O osttra] = Q133 Reaundany = = HOD y joo = 2133x100 Lt Q.ay = 0.9F)RK10O = 33-4AS wnat is mark 2 Howto compttle Abe gacdtent wt (0) Mane P Describe dhe Prewitts and sobat Operctors Aw, Made — The consept aGO mask 16 Aigo lenown at spatial Milesing Mask iS Q 1Ype ag filler Which peytorms Operation aleenty onthe Image: TO APPIY a Pragk os ANIMage, fiHEs MASK IS Moved PoNt fo Point OMANS image Tn dhe ownginat at cach point Ory), filler IS caltttated by USING a preclefined yelationship — oat let Sao = — : | o|-t] Sy= | | 2 }d 72) olojo (jo j-1 jv feo} =i Run {ites over imege dof = Sx f dh - ty@F Oe ay Image Gradient fe “as Ds Sx! dy Sobel operates - — a fo }-1} Lj2it 2 lo |-2] 0 |o jo to [1 -1| 2/A1 Sobal opeseigs the coefticient e& Mowles are acyustable ACtordi ng, to OU Beguirement Provrdect they follow atl PSopesties a¢clewivatire maesks. Prewitt operation ~ BFS Used for edge detection iNan imayze. Prevuitk operate defecis boy dypcs OK edge ie x Hoizoniad sp Vestical ° (Jo Fl ° oOCY What ave eliffecent applications &, inn eae @) Segmentation and also Ue nousn trantforrn te tind OUL Whether the following points (164) 02/3), C44), 044), 'S10) ave tpllincee Or nok - Aus: Irnage Seyrrentotion ts @ funcian Anak dares ‘nage ipond procedure OF OUrpUt The Oe {$s Oro Ov O mmautny with VanOUL Sernents Slyeck clos OF ingiance to With eth DS Irnage Segmentation Tednniques— * Edge - based Segrnentaton Thresnold - based Segrmen*eton * Pesion- based SegrnentoHon. + CUtter- bared Segrnentoton * Watershed ~ Segrnentahen . Given points! ACY) :BC2/3) 10 C3): Blut) E680) Une usin yrorer paaulsatad y. Wich can ce expressed Ab , = axty line Slope points tor) Point Cary) be -aty [tyra “Crs;s) ae b= -B04) (0. 33,1) — ‘Oe ON tess wii eee eee (6,0)plotting ne Slope points (iy) St Ts clear Jhat ca makiraum Pf Bc line Pass Amrugh point (15 ) he, x=s and yes iheve fore , Pinat parr plot Can be plotted as :-