opp ata eg 8 Pca wis ‘<2 DETECTION oF peconTaues ———— 1) Lies nd Toilet test dcontntey mask procening is pei "sponte ofthe mas idee wit mapertcfo ctracen 4.2.1. POINT DETECTION As isolated point is pot whose gry level or incsy sete di its bnckground and which i rated is homogeenot ones ne 4, the area having lhe pitts wih ame intent ere, Pig 4, Pit Detection Mash This mak is wed to detect the sled points doe tate or inriene ‘consists of coeficonts 1” everywhere excep a te cen Go efit ‘The sum of al the eo-eficient i 0, The rsponse of te mask at he cent po the eon i ce where 2, is the neni ofthe pelaae tM i mn be pe cae ey tetce enim en mm SETESTON GF oisconTNuTEs —— ii) Eages Fo idenily these dicontnin ‘2.1. POINT DETECTION A isolated poise point whose ar its background ad which islet evel or itensy is ened es the ac having al ett ih on in eheanenons a yb [ato Gifs Ta] i 4. it Detcn Ma This mak is wed ere lad ply ee ae ee const of otis avi cerpa te es ns neteaen ‘The sum of al the co-efficient is 0, The sponse of the mask! he Cente in the regions oe = yee eee! = Swi vere 2, is the intensity of the pinesTho tela age spain le hangs oe npn scinted with dco eiter he nae itenyo he ft Sa > nage tensity. Disc athe image ies canbe 1. Step dicots 2. Line diseoninitset ges can osu ve te di rocening on Toul ees i rely nig ean | ea ag 4 Fete ie abe bein the acon evi sin meres teran MIST ae Tein a eri ae naan nen eitLz) tor bs tempor geomet pnpery tht it pins nthe dt greatest rate of ehunge off at ean. The magni (net of vetr Vf, dena 8M MG5.9) = mag the vale ofthe ate of change in the dono he din veto, Noe a sited Me 3) a inages ofthe same ic as te rial ated When x an)This operfor canbe ust 1 given below. frst etn 5 mmge oro ig 46 Preis OFEl a PIES F [feta] far wor eTefs] i ii) other effets that intedue puis scotia in ntensity voc, Thus, edge detection sigrtims ate usualy flowed by linking an eb n procedures designed 6 assemble edge pitts ino eatin underes. The following techniques ae used for edge linking and bund Global processing using Hough ransformwhere Eisa positive Kise aire Tp dirccion of t cae (cP saad an divaon in a emi TH re focaton inthe ia rom pelo peel A sil fk : rape each of SP ; a sj x i A simplification pric following 3=P*aciontal scanning proces ia tp 3 rotted ck : o> 4.2.2. REGIONAL PROCESSING The locations of region of intrest in an nage re Known or canbe de This implics that knowledge is valbe reading the regional membership o in the comesponding edge image For these kins of situations, we can uetechniges forking pes ona renal sis, with the desired result being an approximation to the boundary of therein or example, polygonal approximations ar pariculaly atratve because the cx pture the essential shape features of region while keeping the representation of he undary relatively simple Two important requirements must be considered for polygonal ft algorithm such 1, Two starting poins must be spied 2. All the points must be ordered.ree: Mtn ft Pata An elgorithen for find . image. Speci two sani String Vertes fhe pon, Step? :_SpcityatwesndT nd to cmp sacs, OPEN Step3 the pins in P compl cla cane, ne 0 and pu B into OPEN and to CLOSED. I te oc coped to an ope care pu Aint OFEN a in CLOSED. St 4 Compt mene fe in ps on ete (CLOSED tthe as vere in 0 Compute the distanes fo hI whose sequence lacs than bw ‘hot the pot, Vay whe mas sn in sop 40 te asi? ihe veins fos Sep 6 step 5 -° al lines determined by every pir of points Nest find al subsets of pins th re cose optic ines. 2 ~ 0? lines and then pty involves finding nin parisons of every point, to all ines. Therefor ne ian akerative appro to method 1 7 — plane and the general equation of sige we = moa fiom, 7, ~ at, + 5. Inlntcly may ns ps Sois approach involves fidag nn = 1Y2 ~ oles andthe pte (©) (nee ~ 1) 2 ~ compas of rey pol, al ee. Thonn Hough Tanfrm saa leave eros method 433.1. Hough Transom (Consider pot nthe xy ~ pn nd the genera equation of ihe in slope ~ intercept fom, »; = af, + B. Infinitely many lnc pos Sr (py) but theyll sts the eqution = 2+ fore ng values oa line fora fixe psc») nthe parte space tab — pectAn import propery Fg. 418 Din te pl scm tt ee alter, ofthe pant a Dye Dis emi te res ems 99" = 6 590° and DS ‘comers in an imsseAm aprons bse oe Hough Tram flan 1) Obtain any dee inace Specify sbvions int pglane iy Examine he coun of te ecm cel oi pial eens, |) Brame the eeonsip (Example cout) teers pie chosen cl Conny inthis cae uly is tee on computing the aac te sconce pes corespading 1 ive acca cl 'A gap in in sone with a ier all i ged if beat fhe pp ses than specified esolcomes of di crs inigs bho eckground pi © One cbvios way 0 & threshold" tht vepatesthese mods ed an objet poi erste poised = epesation swiss by ssng te ins Pot Hl eget x bceran, deedigo whee he BY aS cach piel fess than the value of se) =4. The nimi ofthe relctnce propa ofthe image 4.4.2. Basic Global and Local Thresholding 4 Titling may be viewed 25a operation that ios ts mass finction To the form © THT ypQr/I)1 heres) the ry level and p (some lcl proper odin schemes compu each pines gray level with #00! oil treshod, This is refed to as Global Thrsbolding. |Trresoldng may be vied = tsi! Gecon cojne st iin be veg ee cen opi i opin i th Sere at groups cad tween la ane ei ney tei ili thie Let (051 son Mx pines and LtS Pam) » Pra) rays re) The thi in flows from te i tat cease we ar dealing ony with vaso Als, Pi th probaly oft Hv, whic i the histogr,p Fially, 4.44 ise to i) se probably of cls, atch we LanTe order to crate where oie the pot! vain ‘And ois the between cs varie, diets Gh ~ Pom-Pom “This expression ante writen acs = FR(Mn*, by eluting Ba, 4.16 ab 44.4, MULTIPLE THRESHOLDS, Mull tesla be extended an wien numberof thresh, eam the sepabity measur o which tise ao extends oon hen ne lees il, cae. oC K plana, Ceca etme cee | seneralizes othe expression 04 = 3 rim-map a‘he lowing relationships hold44.5. Variable thresholding uniform iumisate play & majo ting so subdivde an image ing One ofthe simplest approaches resto novrlpping Rectangles. This proach is used to compensate for eflectane. The Rectangles are chosen small snough so thatthe iluninaton fea ony AND > bm Qo, M) ~ | fase atemiceFi. 413 () Tet image orate by (u's method (9) Rea Thnesholdng based on moving averap are smal with respecte image Example: mages of typed or bandwrite ext2 Resin Spliting and merging 4.5.1. REGION GROWING Region growing ie aprocedue tat cup piel or sb epons nc te folowing eps based on predetined eit for growth I © Ao inal st of sl areas is iteratively merged secon ‘Sart by choosing an ehirary sed pinel and compare it wih rele ‘Region is grown fom the see piel by ang in neighboring similar, inreasng the sie ofthe ein ‘® When the growth of one region sops we simply choose anak ch does not yet belong any ron and wat again ‘© This whole proces is continued util all pines belong to some bottom up method Region owing slgortm, Les fs, 9) denote a nwa ay, S (5) dene seed aay contig 1S: locations of eed points mu 0S ewer ancl Q denote a predicts tof4 Problems can occurif te (athens el pot son To counler the above robles, simtnous yoo poe tpen developed % Similarities of mihhring gs ae eo uni owed completly dni be rece & Nosinglerevioa Jpnedin gow tsa tine pas ro expanding regions. © Amumberofres ons wil stl Similar cove? ie pene. parallel compu:© describe the spliting of the image using a tee sevcnre ‘ee tht is wes in which each node as exaetly four dexcendans The images coresponding to the nods of «qua tree sometines ‘pions or quad images