ECE- | SIGNAL PROCESSING FOR IMAGE AND TL | ea VIDEO N i t rg |c) 1. Fundamentals of Image Processing: Basic Steps of Image Processing System Sampling and Quantization of an Image — Basic Relationship Between Pixels Image Transforms: 2—S Discrete Fourier Transform, Discrete Cosine Transform (DCT), Discrete Wavelet Transforms. 2. Image Processing Techniques: Image Enhancement: Spatial Domain Methods: Histogram Processing, Fundamentals of Spatial Filtering, Smoothing Spatial Fitters, Sharpening Spatial Fitters Frequency Domain Methods: Basics of Filtering in Frequency Domain, Image Smoothing, Image Sharpening, Selective Filtering Image Segmentation: Segmentation Concepts, Pant, Line and Edge Detection, Thresholding, Region Based Segmentation. 3. Image Compression: Image Compression Fundamentals—Coding Redundancy, Spatial and Temporal Redundancy. Compression Models: Lossy and Lossless, Huffmann Coding, Arithmetic Coding, LZW Coding. Run Length Coding, Bit Plane Coding, Transform Coding, Predictive Coding , Wavelet Coding, JPEG Standards. 4. Basic Steps of Video Processing: Analog video, Digital Video, Time varying Image Formation Models ; 3D Motion Models, Geometric Image formation, Photometric Image Formation, Sampling of Video Signals, Filtering Operations. 5. 2-D Motion Estimation: Optical Flow, General Methodologies, Pixel Based Motion Estimation, Block Matching, Algorithm, Mesh Based Motion Estimation, Global Motion Estimation, Region Based Motion Estimation, Multi Resolution Motion Estimation. Waveform Based Coding, Block Based Transform Coding. Predictive Coding. Application of Motion Estimation in Video Coding Text/Reference Books: ; son R. C. Gonzalez and R E Woods, Digital Image Processing, Pearson Education, 2002 ; = 7 ena > AK Jain, Fundamentals of Digital Image Processing, Pearson Education,1989 3. W Pratt, Digital Image Processing, Wiley, 2001 4. Al Bovik, Handbook of Image and Video, Academic Press, 2000 5. Keith Jack, Video Demystified, LLH, 2001Description Page No. 3=9-20 _|Unit-1 Fundamentals of Image Processing | on 3-94-20 : Ss 4-4-20 [Baad oe se, — ee a eet — 19-10-20 |: 2-D_sudundancies UM =20 Thuory of Waele ,Disertts avelid Transfeem | 72, | 23-11-26 |D T —rr ey AVNIT! Fundamentals of. Image ‘Troessing \ > es Tinoge P u. jess ‘ meowsing ~ Thi field of ole im [9 Cs ing show to pus wing digital ‘ages by eles af o digital compuitin . Digital Tmagt- A cligilal image it a 2-Dimenrional funchon POG). hou X and Y au spatial coondinaitas, The anpliteds of Pit called dntuwty or gray luvel at dw point Cx, 4). DIP wes dhe fotbowsing pro coos: ‘ * Loto = bevel procs — inputs and owlputi au images mid-bvel processes - outputs va attributs extraded prom tapuk : high Lovet preustit an onsembl of sucogrition ef, individual ede” Piak - The Amollest adement of a oligital imoge. S Baten alument ee “e A imple image formation moc POX¥)= VC Q) * (x 9 F(xy4) : intensity at Due point xy) Loup Whumi notion o< Ely yew roy)? raftdane /drorsmibsivity — Ocnbyyed Types of comp uitvu zeal prowau Ccv) ws: T Low level prowssing — Te dwotves puri mibive operations Auch ob image prr-poo arg do sudut newt, rortrast enhanument, Sand image, Aroxperge « : TE Mid = Javel procmsing “Ts involves ctasks auch as 149 mentation , dextxip'tipn of, thest objeds do rzclua thom dea form Auitable ton computin prowsing anol classification of inclivi cual objects ., D pighen- wd pocusing- a invotues “making denis? of an uuemb le qrucognized objects a4 in tmage analysis , Performing cognitive funchians normally associated sith vision,"nomadic ight Lands! 042 do 0714 um Radianu- otal amount of ervgy ee Luminany~ Clr) die amount of trrgy on cbseuen preci ve, OO a Light Aowu. i Brightnenr— a subjective dewwuiption of, Light perception that ia impossible to masumt. $t embod Wu athsumasic notion of, intensities Image Acquisition - iii Fito, te Linage Lompling and Quantization : Aa Be A Peon eee 4 Be ee i . é image indkeraihy phe sampling quanki zahion Continuous A dean baw Digirty He Digs rg vnogy Prom AJOBin — Coondinats valuag — armpiTudls valet Hy womFinuoud imagey-q-2w 6 Repouserrting digit ob ‘enoges - An image parent awa rd Flxy)= PC 3) , PCL). PG,n) aoe aA A Fi Flap) -» Psi) pivel 0H Ke mtenben oh las Tae Ces) #004, = F048) Pollowing Ormray iu a rupseumtabion in pono ae A Man MATLAB, Total novel bus lun an annoy b= MANKK Dis oul intensity jatewal Co,.-1] stata bas : Binoy ‘image ony tase Ahades- Black ancl Whitt Greoy nag ~ Mo du eee re ane uid; Spatial ete an te yusaurd of the amablut! dis wurihle ALEoil ‘uy on ‘en fdboded with Mints poivu pur uuvik distane y ced olistana , dot por ink (dpi) > tmynl Toten a risdubion— gt is dhe smallest, dent fost intensity + Stu stated with “eb, rabits , Tnterpotation- Stir tu process of, uring Jenewn data to utimate tnknsam Values . 4. Zooming, Ahvinking » rotating . and groretbut terrection Ub in domitimts called ye dam pling. Tr Ls an im Smetrod do invuasw (or disease Se number 4 pixels in a digital image. Thon ara Hout inturpotaton techniques — iy Newast vuighbour intexpetation. — Ue umsicow onty on ntighboun do interpotati prot ‘inkormak on Ht) Bibiwon interpotation— Th dake in eniidlnabon 4 naighbouna Vay) = axt bur cxytd.. For one \mage, Hu coehtici ents Abed an Ind Bo mn nals . iy Bicwbic Wntexp hn! a takes 16 neighbours in carricleration Ta du bot nen the Hout. Basic sulolionships fetwun pixels - A. Nughos | nigh boos o} pis P ot wondindtes x9) * Na Core Amgen LPS Cat yy slate shy, by yedA Ny (pd! 4 diagonal nvighbouns oh PE Orly) Can, aid, 1 | Cxaty-r), Cosh yD : NgCd: & niighboion dp» NaCpY= NyldUN 6) Nol ny | Ne | Nal PLN, 2 Adjaun = fe V bean aut of Interity volun, We Nye + U-adjaciny + Two pixel p and q with valu from Vv Mu Y-adjaunt if q is in te att Ny (pd. + B-adjauncy + Tio pile p amd wth values 4 B-odjocent Yq u inthe mt Neo. V one sm-adjauna? Tie pin p and gq with valu from V ov m- adjaunt oh bq i in the ate Ny Opy 9 O& Slt i> og is cy Ut at MAG and ack Ny (DA NYC? Ras * no pra whos valu au Ipom V. é 4 Te tannealivily «Tae pinche p ancla, wih valle’ fiak'y ° YW connected ig UL prom sthe aut Ny op) _terenecbivity Po] EV « pogo B-com necked if 4 ® fom Nip. * MoLonre duly § pig Ev, 4 ene Dae MoD ond Ny@®Nny CQ) =. (9 amd (13) osu m—donnectid diut Jo first tend Stowtaatiy 2) amt 62 coe my -umneded CD ond Ue) Ou not m- be it viata i a 2 tonne cled cee Osu ime 7 age Mrtomeded ssa, aeeond emndition Pq au’ m-esnnectd ip Grr and (3,3)< Iwo image subseti Sand & ES, “such that gp aul ej: L aw Lt on Mradjacnt Or Rroctyaunt, #6 Bord Gam adljaunt 7 & ei 4. Path- A path prom Pliny) dy Glib) ina agquinu of MUstindk plabe. Oyyds Cer Yyo y- Ctay Yad or (4 = GG) and Cap 5 Yad = (St) . Gee Yid BH aApaceuct ay Oy yin) Pox Coon +o Length of thr pabt - TP Hey Yo) = Has Yo?) y dhe path ls a dosed path. We can define 4-4 8, and m-paths Drastd on they Tyre of adjacenay) sued. ts Wend pear} t Cxaniph ? Adjaunuy. and Path. Grsidee v= C4, 24 On tan bys 0 \---O|- d 1--@ O21 22 Om db RO 0 2 0 03 Osa Iya oo @ +a 8- adj aunt mb oolji The 8- path from U3) te (3,3): DD C9 ACD (2)2) > 03,35 id (QV AD— CD Th m-path from CVD WED! WD, Chad, 2,,6,3) 5. Connectivity- fur S rupert a Aulatop pixele vv an image. Two pixels pp with wondin ate Cr. yd andg vith tovndinates Cra, Ya) ars Sard to Le cannented in 8 oe L etists a Path Ge gods OH Hs Om yd vi, Osisn, OUYDES, ‘ 9 14-25 S supreent a Aubset sf pivels ° an imagen \ © > Sor wey fia p un S, sty AU of Pratls in S anet 4 tomedid do pu callud a ceanedhadl fampaness tt s > 7p S dar enly orm conne cid Component, than 4 at, Connectid Set, i / cea “Rog sugion of dar Image iP Rw cs conned , > Ter sugiom Ri and R au Aaid to adjacent ip y* Union fount a Connectid att. — > Region dhat aut not vo be adyacent aye paid te be distoig S. Boundary - (— Revduc).— Th Bouncl any of AUgton Ria dhe auth pinels iy dha Jugion’ Uhat have cu ot mosu Migdons het aw not in R. Tf R happens do be an wALDu ima, shen Uh boundary as defined Os the nek ot Pile in tty fout ad fost sows and cotumins 4, 7 imog 6 Fonpround ond Background - An image tonteins K disjoint Jugionas Rey kal 2). -k, WR denote tha Wren of all tu k region ; and Let Rio dnote ih Al sth poisls “aR, 4 foruground pou ‘wo Ru uw cathd fo. und . AU tM ports tn RuS ia cabled Back ground. istana Measwur —=ems Ciiven pinels Pr 4 and 2 wth coondinates Cy yds Git) ond (uv) rerpectivelay , the distanu funchon TD har pelleot} Meeks ty Depa Eo DD O.q)= OGD paq MHDEV= DGD DOp2) << Doig) +DCq 72) Ww)a) fsuclicltan Distenu : 10 Detpgd = Sage yo b> City Blok Distanu ! z Pulegre leads Tyee : | 2 Cus Board Distane : az = Dy Cp.q) = max Chest, tye) 2. 2 | Cty block uu boand St Mashematicd Operation in DIP: As {a Oo, R=foy by | On Oa bar ba AB = [aub, dubs | ATB=Pabjra.b abardub. | ay Mau: Sah Oy by, ar baa Tesla nbs +4rb, Ay bata He H[FG 8] = 4.009 HE a Fp + OP Og y= Hla kody + Taser) | = A HNOD] + ARO] = a gibs+ a QiGuy | Additivi Homay ersity His said do a Lintax operaten. His said de be a nonedincan oprraten if it dees not mut the above quah on Anithmetic oprrakons ativan images au adnay cperahens § Sum epinakon ACY) = Paap +g diftouna operabion d Gr, Y= Peap - gy) product epenaton pluyy= Fly 4) x4 (7,4) | division operation vy) = Pony) + 9,644) 342 Image ey Adolit on of noisy images Pon noise suduchon & : Men GgLNG | a ag Noiselers ‘image . Alyy) > nove | AE ety pain of wondinates Cx, 4), the nol is unconneleted ancl har Zero average Valu. gy worvupted image gm es Pa, ny, yD- M Reducing tht neat by adding’ a aut of noiny images , fais) I= eS gu) E Laon) = € [k Egos] JE [4 é Cierop 4 aay} = Paps eltencp] eo = 4049) : . 7 : . eae _ =| Varionu cb the > SF yw = & Exed= LE asta EHD Tinagt Subtraction ¢ Mask hlxyd—> an Xonoy ineage ef a dugion of a pakinntis ody. Live image $Gyp> xe ie caphoud ot Ty nalis offen injection of ont conrad madison ; eg ee Ths proudunu giver a movie showing how th contrast medium Pepagats diawugh Jht variout arteries vin tr aria Ling bsowel Set and Logical Operdlions ~ dt Av be de element of a gn A= Cyy,2 J2= £05495 © Complement of A Ke rX-1; j- Atala image . = AS 1, 4 K-29 yz ea} Keo Intensity Lit per pix’ : x ° a ca theo ry Ack images + Av B= {may (as) [aed beat i: bee Not A yf A AND A xok B und todky™ (ror 6)Bipatial Operations - . epee Fae tren vedi E ° Bingte-piut opuatiom.- To alter die vals ofan imag: pists rand on sia intensity, mapping Punch on ia umd. $ by S=TED ae AT obtain a ngative ef an S=Li-2 Q- bit image. , oe + Ndghbowthoed oprah: - Tu vals of, 2 pial Gry) is tet ovenogs volur ot he pile in ruchangte. Say cusitnd ok iu y) . This ds a kind of LPF . Et wilt kup thy owe frequiney components - image F smage g © Gteemetnic Trans formation C rubber- ah trans formation) -~ Cop = T CMe 5 4 © Alfie dram formation i (xy tp = Cv w £ [s on °| eg. Tdentity 1 Scaling ; notakon, ty tae 4 Pranslatoned, shar Cverticel, horizental) , R ing | Rota Translation Shen aay Taentity Seating aan mentee” Gerke) Chonizenty 1&0 oo bol \ j wEv x2 venO-wsind/x=v4d, | %=v4Syw meV | 007) Tin® sind 0) ft oo] fr oo Afbine a 0 ©| |sind «x8 0 o Lo is" of}. Mave i[o 5 stloostto o 'f |e t afiloos Go rdindle | %V Guabions! Yew | yeG yavenOrusind YoWety Yaw yesvew : er | SN Exanpk 5 Ps Fr he ' | ' % *oo Transformation Panchen. *> inpuk pixel. . Bina, tur wu claling with oligutal quartitir, volun of the trans formalin funchion atypically art Atinad cw a one-dimensional Way . > da P= NPER. Wo Dram fonmation ke £—> input image. - Pe Mx inary id MN xl v powig, R=eMxN ——> MNxL roist > A MN XL 5 = 4) . id is Nx Mnxe NI N nye a. MI NA Tams 5 Ef Posy Cm 44? uso Yeo yp s inpud image forward trameformabion kennah vo transform variable. v= O),-.M4 &V50).. NA Kucrls (masks, filtoa, demplatir on windaudy ean Lt an (ur in which the Values ef Ake (mask ) coef picieuts datenmine dhe natwu of, the pow uss Leg. image Ahan pening ) Agri wb equation Mey Net Term We an alto have Plas & S_ Tluy) s %Y, uv) U=d veo rit Tharifponmahen Kernel analysis equation 1429-20 Forward TransPormation Kexnal Jt said to bx atpanabls iP ACY UN) = HCHV aa Cp). ; = oes Tt proviols sos computational complext ty= SoA oy iy laa TB Hh Cis Pitan Squad to mn Cyv) Tee sete ys UWI Iu Ce Wd oy Cy) “ Tue Dimensionel Paunion Tras farm - *luysuvr = = 2 We + uy) 7% Y 2 Apacial voriaby Ly > Grgusney vasiabl tr x,y dirscion. reve kerul A Ca ysun) = 1 ain sees) Tuy) = . =! ti, pe OED > feo M-! : fogs oS ST yyy elt CRD elec 204) | Fran bom f Liew] arom 1 9) i transform | | epadiad re pirral approach be aperating i bn Ha Lintad Hravsfer demain Assumptions + forusad and invert Kewl ane Aymmébd ‘od aby. » + Pegy) aa Aqua image emxm) hur wt hae’: T= AFA on T= A, FooA nim Mem Tg SMM & Porwand trans fanmation aan ae Kemah in separable form To obtain ahi inioae Transform ‘ f= BTB= BAFAB maven tranahenration Hom | if B= A => F= aap as = TET =F tan ao uae wall gab an append madien £2 BAF AB Probabilishte Methods — ft Z ,1=01,2)..,L-1, dante Value of all possibly IMINO In om Mint digital imagYr peolability 4 p C2, ef intemily Lowel 24 occuring in 16 a givin imap in. estimated as PCr) = Pe MN 0, no. of tims dhat tensity Zp cco in He image MDP Total no. 4, pubs mi =p Seanr-t ul eT man Cawtneye) intensity is given by m= 2 2, ple) =o ° “Ie : coy , ul “ The Vodtanc ef, She intensities is given by ae = tam) pd + Th on moment of tee inkensity voabl 2 is Un C2) = = Caer p 2) High vorianu means high corals coritrast i mage (1-4-20 San pling | Theosam « * - he a p Deh po rede tee ee pele et oN ear! rf finale PB Cmbx ndyd ond Aw = 0 fee [Wl = Umax (VIZ Vinay : Ae es Von 2uUmox 2 Vinan Jean Baptiste “Joseph Founien Atated, that any pewodic function Cam Le expouss eck a othe Aum of ines br cerinas of Co flurend Prguenun, * Nr peviod:e finite Parton eam te expand ts Shia Aum eb Cablecl on Fowrier Aenies - Non- peried firide frunéliors com Le eeprantd at dhe integral of airn and/or worinrs mulk pled ly a weighting funchon. Jatsford clion ws cald . Foureive Traraform. 4.261 ID Fowir Tram Porm and uf invent — Ine Runive dram form , Pl) ofa Aogle voriabb continuous furction Pex) at Fy = f Poo dy a and invour Powrin ronform Pod= $f Feuy ofa - Yao varias ¢ Funds ff ge ITO DP tay Leng FP cea i OP dud. Sourivr Tram fonm in dlisvute. domain aways sxist : eet -jurus, Fw) = 2 = rr On) 2y-— A=} Disoute Fowsin Trans form (DFT) TORT? poos EF! Feu SY per me ya MI aa Hou Eutin's formula el = cos 4) hin ae Fouad Trans form cam abo be suprusented in polar coordinates! — pryy es [E¢ay| eI PD [Fel = Pree + zw 1% (w= ton" ear) oe a. RO) Foun dpectium / Power Apsdtiot dunsity Pla) =1RC) Plud = Rw + Tw) oe amt Panction:. Poo= Brack (1) <> FOD= ataind #7) | a ceded ob vuco 3K sowie ag ee : lush dy ‘plyinf P00 ram u i mack henlen ee! C-D™ box tabig explained on P19) |18 422 Uh. Top Dimenstonal DET and di Davee dh 2D DET op. fPunchon Cimage) Poyud, of sie MAN i - MA net ys > ds given be payee 1 Mt 20 Cua YY givin by Feu) = 1 5%, Pow et “aot oud ve HME Oy MEL OAj oe NAV IDET 2 Poy 2 MIN jena va, P= FF Fade (8 a ee Ola tore Yeon Ne a Apakiol Zi moge Vou ables uy > fq Tron fem / Byeeny wanialles Fourie, Apdrume lEcyvl= £ RC) + TG] + 2D phaw longs bCuv)= ZECuv) = Law Env P pha ange «PCy es 2-D_powtr spucbum PEvr= [FCS Rav) + uv) Uo dram form or Ahipt the fouovor apscbum Jo the middle C4 PAY, dhe fundion 4 multintied dy nerd “'¥ Teaap ens] = FC, vey PCx,§) 4 Feu) =| 7A al My < : z x cy My ne he sty fu) Fem GV : a S$ 6 WN, STS Ma ‘ ws mae Ne Foo)=1 5.5 top auras of, devel of Be ethane a aaa Ge Gemponent of spstirum Can goin preguancies on 0 at oxgi)2-D impulse’ drain dy fourier Fram-form 4 . pg Sarat? - H 1 sar * undurs aennled ee ane eee teaial Faw) ros | (the™ Wel <2 <2 o athesusie i Fluy)= i ta Bry) Ae A an : SS ATE: sine (nut) dine V2) 9 “Pro poiti ; D kaye FQ): 2 xt eo [xi] . | A => 260 €-1Ne> Flu) aby ee veer 7% | 4 7 iM x(n) FO0 ER) Cy Etuu) = [xd] & Pramlahon ments rly » Conjugate Agony 2 > Gown transform ofpe ral function. Heap. ab b canjeget pment ie iP Poy) nual hon, Flu) = E*Cueev) Trai = (eew-v9I be. apritum of Pourion. Drangtorin’ ts 2ymmeubic Ob POP is iomapionrys na F Cay). - F*Cu-v) © Ply yd cA OM) > Fur, vevi) teed) “4 ” ree wh We and Ue Sti eC aH) 2 IYP FoDeEN™) oy Fam * £lx-%, Y-yd <— Fae Sinitoly toMy 8) > FOC” « Disthuibutive and Aualingt F Pha. nou]. = Fate] + -F [Reg] FL Aaa kop] # ZLAal. 3 hoa] a $44) > a Fo) F (on, by) > ar Feu, or "Rotation? x2 x00, yessind us wisi vz waind Fx} Y) Become 26,8) od “FCUv) Bteoms F Cu,b) £01,040) <> F Cw, O42.) tee rotary $ by Oo Potetu F by Ox and yin voa. . Penio icity 2 Fous)= F (ur Mv) = ECuveN)— F Cus Mv ND ao Flue Pla My) = PO, yen = Poa M, ya NY 184-70 | E = t : Tir Ln nya s Povusand Biase form lannd ; NA wy F(x) = Ss Te) SHY) Convict tam vice BOW. Thrangh © hehe BU oF & Gs § Bp O27” whe “Wey = Fee . eM s - ! i 6- int O-w'0=40 Teel tt =D ZLB wr orthegonakWe, Wis War” 21 al 6B k= OB Q Aut of none zo, vedlou On imubsally / pave wods outtis'gen = < Wig KW, oF UW Wy? 4.04, < Wes 2 +... > F< WELW oo Cw, a FOFOA LL FO CU WLr HOH. Sear Bis we wir Swine? Pe thay ave ieee eee S Wig Wiz = cwrye> ) 2 cothbiient” aa vote Nol . Aimiloty, Poo Ti Ted, Sexy —O. . . SS catfficiuas’ 400) = M0) $ C40) + TON SED... TON) SCX, WAN =D Teoye < $.04,0),, 260) Sens a| Tews < SCyw 00 > MATRIX REPRESENTATION } Pe (roy) (> te [tw 7 e >. fPay || i fm fox] From ©, wt howe mage. Ranchionene eee icc oe ous fox WH OND, NP DD Su, framfonm Sc) a > TT Say Kool Sv) Suwa Now Tis < S Caw , FOYD =p Tare 2 PP > ¥ GOA) NI Gmsidin a matux A, x ‘ : 7 Aa[cl | = (S00 Son Save Sowa} 2 C8 8 -~ Send se So Sy Syn Syn Souci Snape Sry Sear Sram tum t= [ ef" | [tle <8) 417, Soci bay FTL > fem) Now AAs [ST] [RS SoS) = SEER Soh? ca ee LE SiS SS, ST Set Sua. a Te 8h Suu NA Spo Suet ray On Ror, Oy Sen vji ° o o 4 ° als qu orthonormal : => an Was ans WA AA" Lo oo 4 7 >feear] [Feet — Uv We amumed dh No dram form bas yedoru Su fer defo, wi] + ap the strane formakion rate ars raat and orthonormal ab <6, 5 >= SS = ie: aa ; _ Lt, kek“ae gaint appewack cam de ‘whended ‘to 2-Ty Me from page and 18 wt have! : Roan Cuysuy) Thy) = = Peng) SEE Tu Sng wo! , Dhe draws feownation Kouvl aos repanodt , than abyy unde MCB In CV. od dah funvion. Kip Ahonsfornat “nat ee fo ation kuuwls as mal and ortho rexmal fee both of L au Arporobls ad pumm ebuic thon * FA Jo [Fe atta [ : Okt poet awd» ro e bi " cues ud = cubilieg F t a Ane wees omplix rvelutd Labia vectors an ortho re ) + i md oy PSS, So =< a Sp = : = sets » [rH AFA Jad [Fs [peut Aen ae -@ ho AAS AAT. RATS WA aL aa dhe ; Cc . _ FP emma pending equaton an: [CAP Fe a . 4 — a 2 (W) SED ~-Real and erthononmal ~ ols 5 Moc Gevers) Gries me enionotenes ° © %® ap * Rat and eel On oes eras / * Complax, oT gitl-o 0 o et wa Syoroin Cquasion Fors 5 2 TM) “sud ar) Fass ttuySuySuy= S%uv) Sloljuy) =. SCON-Luy) ay SO,ouv) SUyuwd ee SUM urd S(NA puy) SO) - S(N-AN- wv) MioUvt 42.00, ND who Cog) aoe ate x. Fi com de tepund os a dintan combinalion of Ni mabaus of ate NXN hat in Say fer Ue O, Wen! . Sh SCaysun) au sad, vedas atparable and aymmers Suy= Su E> digital Trays . Suv> Basu oe cam Be ascranged in an NIKTD mats. dhe Disoute Guhe “Trans Poon (2en)- DCT st obdaind ty ull tating he inves Branstonm Kennel.” > : © S Cx,a) = ck Ca) ews (@xpat) : Kw =4 ik focudol te aay TE Poe ue Bye Nel No : A® (sy Se Fe Sut] Now ce Ap 2 anabysia quaten Now othe Tian form ahion fonre) da arpanabls , we get Sry ude «Cu cv)’ co((meadut ) oo (@ air) 2M 8, Com) = occur or Casi) S acy) = xO) c= (Coys) IN SCxyu,v) i real, hep wrabl, ayenmative & oxthononmal Keynsl SOnw)= SCL) S$, Cy)—— Op pet? T= AFM ad’ FsA TA fism 4 @* comprdahm of DoT from DFT N= pout poucdi dity. 2 ~ Pount speniodi d ty A> guaid Zminimiu vee Loundang lis continuity . Det of N-poor Bandin POO cam Le Compuled from JM DET eh aN~ point dymmibvicelly wetendid veucen of fo, DET assumes DCT assum Te toy orden Step t= Aymmébucally exctund N- post dis out ferdion £00 Jo ottein gho-{ 100 fy osxeN-! £On-x-) fer NSXSQN-7 ey Fo = KA gu = _ > N+ 2 ° ee 2 r f= [For ey... RNY] Bel qe gavnd Sepi- Gompuke thy 2N- point DET sf Fusing @@ . comer and oxtho nowinat kund. | 7 a ; OR: ae apa : Ap > Diana form abion motu DEF titan t 9 -2B 2N DFT StP~ ge Qo dement cobumn vedox R= Lhlo HO heval” , + whew hore eV uso. Na and beh SG ine] 7 adamant oe Steel — Det a Punt ches Ref. Gohet oF rk Soe ‘wut Pi codon [Fa dorand Besdbict |— Cdlone by self) Examph WO Compute DCT ¢ for X= 0,1, 2,3, 1. 9 = fm. Gan P= [Oo 1 4 4] Lex , OS*kSN-1 FON-x-1 Ns%e2n-t 2 Tram foxmaten matiix = Asli | ' ! a eg oy t 2 ay 4. DeT 4, =Re {Sehety} eRe{ [orl 1 ot toto @ 4 Je-t -w j = to-j -1 Jes w -l je -i- (ee! ee 2 28 . ~4j- Yo -4 Heo IN Hay = HjOFY +109 HUjeo~ 5 ° ‘ — 09 Uy +GaAH4os GHA D+ AFH 1 JOE 89 1-H =] no em w tally-Sin My) Certtig - Si 3% i ie bar-(0a8] | ~¢.18-52-S¢ On" jou 7-10-20 26 -D bun Hon fOd=x* : g=[o We a ' a ju -e"e 4 4 4 4 Pa 1 6 od pr 28 49 6O3-4)~3--9) | -A-O uedy ie ve Va City) © jD4j-9 6-4 3G >| ve I f 44 OB = 256 Mi ajnay L-Ol8 joy } =| % 7 S:fow) 082 50°38 h Orl- jo ! O8- 0-92 | i) 13 | Pale yeyy OB —}092} |-o1e—~jo.4y| DeT = 4 |= 6-66 -0.4RSputiat Domain—- jmage plane ubdlf, diructly procs the intensity volus of Lhe image plant Trassfonm aemain = proiss hy rawns foren cothbi dents, not dinacthy procs the intensity valu of the image plant. Spatiol clomain Proum? §— 9(xy)= TE Pr yi]. Cwpat “image ee y & opub image origin ___» an oprredon on f alafined over 4 neighborhood P ays i ata of point Gey) 3Y.3 ntig hbysrlood | 3x nig Pre x 3.2 ome Basic intensity Trans formation functions 0) Contrast sbutching function by Trusholding funckion 7 get) Be et eee ba des trnage nag atives SLI gy 4) Log Trans fosmahons | Donkek Lights | &=¢ fog CL tx) | ae © Power - lo (Gamma shane bovrraton S= cx ouput intensity lavel 5 vr a ty Ey Tape intensity Quel, v12) cenbrast sbetihing gd et a (a ny f «¢ s E CY I: ee eee ° a : ERu pero oA 8 tl um Rghdi bh Unkensity highlights rawge [4,6] a x Expands de Seong of inkernity evel in an mags 40 thot Ut Spams dy full ikea ty meungs oh the eatery medium ex 'splay dui... 5 é 4 hy Bik plane Aadicin aire Bem Trates d ¢ highb girding Pou indeusity deol Hang, ue coud Dightighh Ha 3 uxthi bubjon made by apedtie rts "bp dota) imag. anyon nu, pang LA,C] aad sudluc al otha i nderuitis +o @ bower fuel aud po vies all othen indemnity levels . Git plane & (188) oe Histogram Prowming at 4, dene she intensiha of am L- lel digital image Pry) for K2 01,2)... LL. Te wanonmatind histogram Pow dibinwd as Hide for KeO'2,-0-,L-4 . NL > number of piu in P with intemity ry, Subdivision cf Hu intenity saab au “callid histogram bin. Nowmalizd histogram of P is dilind a ts re hUw) 2 Oy Ma news poke PC. aoe oe a ee Prasat PHP. Ba Pam aoa aimply cabted nistoghams on image istograms « M Bots also ul paren min SE pd= 4 4 a [spar my | J hi potbabiity density Bandion 7 ths nandem variable Hy orasaseg>\ 2429-29 go “Histogram Equalization — To inkenity duvds in an image mary Ibe ofeued os nand’om variables in Hu dnterval Co,t-1]. 2A Pa (rd) and pis) danot Ye porwbability duraity fanch’on ( p po + por) of random variabls mn ond A Prod is & PDE ha Bry) Ro, mw contrast image hawing anbitrany POF. fs) daa PDF ake high cortiest i how wiforn De S=TOO, bene Ly X29 Trput frnaga, inberaities SS Oubpub Tmagh Unbenaiti Condition tor 5 put Trash intensitien OG TIm) ip mombtoiealdy inouaring furdien in Jkt indiavad Osnell and by osTOD SU fr Of eL alse A= T"Csy oor Lie monotonically inovaring Striclly mero # Tawour magping is ponies Panetion sonically cans for sdbuitly menobenicalty invaasing fanch'o, & tut Bld omd RG) dancte Je PDF: of interities of, x amd» ce dum difinent images. TF Bby ond Tr) orm Rneum and Tors Conbinusus and ciffirertiabt: over Jha range of vaduun of intoust hen, J PDE df He dans formed Cerapped ) vordelt s cam ke t 1 writen at RGd= 6 oo [gh = Bibs im nckion eh paraatan importanae Un image A drone form: . poo casing 2s Setime (LN f Bytes) © equalizer function i A oD ad dt is Hy ans umclen Hy Canve Noo if yt p& with obi ss JPome Lt ahem S2(tD FR lade e(teDx te bb Go, tm aloe wo cenditions art satisfied