20.500.12556/DKUM-14808 ANALIZIRANJE IN POPRAVLJANJE KVALITETE DIGITALNIH SLIK, ZAJETIH S FOTOAPARATI MOBILNIH TELEFONOV ANALYSIS AND QUALITY CORRECTION OF DIGITAL IMAGES ACQUIRED BY CAMERAS OF MOBILE PHONES Pri zajemanju slik s fotofoni (tj. krajši izraz za mobilni telefon z digitalnim fotoaparatom) se pogosto srečujemo s problemom prisotnosti šuma in drugih nepravilnosti v digitalnih slikah. Vzrok za slabšo kvaliteto zajetih slik je predvsem v tem, da fotofoni nimajo vgrajenih kvalitetnih leč in stabilizatorjev slik. V tem diplomskem delu smo se ukvarjali s problemom, kako izboljšati kvaliteto že zajetih digitalnih slik slabše kakovosti. Preučili smo različne metode za popravljanje kvalitete slik, in sicer metodo Lucy—Richardson, regularizirano obnavljanje, Wienerov filter in slepo dekonvolucijo. Na podlagi pridobljenega znanja smo razvili lasten algoritem PSFCALC, ki popravljanje kvalitete slik temelji na izračunavanju enotinega odziva slikovnega sistema (tj. funkcije PSF). Vse preučene in razvite metode smo preizkusili na testnih slikah realnih poslovnih vizitk, zajetih s fotofoni. Uspešnost popravljanja kvalitete smo kvantitativno izmerili z opazovanjem števila prepoznanih znakov na sliki. Rezultati so pokazali, da je naša metoda najuspešnejša, saj smo na testnih slikah, ki so bile izboljšane z našo metodo PSFCALC, v povprečju razpoznali največ znakov. By capturing images with photophones (i.e., shorter term for mobile phone with digital photo camera) we often encounter on the problem of noise presence and other irregularities in digital images. A reason for lower quality of acquired images is that photophones don’t have build in quality lenses and image stabilizers. We dealt with the problem, how to correct the quality of acquired digital images of low quality, in this diploma thesis. We studied several methods for quality correction of digital images, namely the method Lucy–Richardson, regularized restoration, Wiener filter and blind deconvolution. On the basis of the obtained knowledge we developed our own algorithm PSFCALC, which bases the correction of image quality on a calculation of imaging system unit response (i.e., function PSF). We tested all studied and developed methods on testing images of real business cards acquired by photophones. We quantitatively measured the effectiveness of image quality correction by observing a number of correctly recognized characters in the image. Results pointed out that our method PSFCALC is the most successful; namely we recognized on average the most characters on testing images corrected by our method PSFCALC. predobdelava digitalnih slik zamegljene slike popravljanje kvalitete digitalnih slik mobilni telefon fotoaparat optično razpoznavanje znakov digital images preprocessing blurred images quality correction of digital images mobile phone photo camera optical character recognition true true false [S. Maðarić] Slovenski jezik Angleški jezik Diplomsko delo 2010-07-08 16:36:53 2010-07-29 13:31:17 2022-04-12 22:20:13 2023-12-18 03:19:28 2010 0 Maribor 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.932:621.395.721.5(043.2) 14295830 URN:SI:UM:DK:HVOPUZ8N VS_Madaric_Saso_2010.pdf VS_Madaric_Saso_2010.pdf 1 487BCD55138111EA077409A15D02D648 9b10ea01150e4c8b84f8bea1b8d9064ee2920a6316f2bb94a7d2b54a6ef0d109 3dfc3a26-7c04-11eb-bb7a-00155d0001ca 20.500.12556/dkum/fc1e99e8-1944-4f34-b995-7c880cdc50ac https://dk.um.si/Dokument.php?lang=slv&id=16223 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0