20.500.12556/DKUM-70636 Optimizacija slikovnih operatorjev za segmentacijo mikroskopskih slik Optimization of image operators for microscopic image segmentation Delo obravnava obdelavo mikroskopskih slik celic oz. tkiv na podlagi večstopenjske digitalne obdelave slik. Namen je bila optimizacija morfoloških operatorjev za segmentacijo mikroskopskih slik. V delu sta uporabljena Bayesova segmentacija in genetski algoritem pri iskanju dobrega kromosoma za optimiziranje rezultatov slikovne segmentacije. Pri učenju in testiranju smo uporabili slike modro in rjavo obarvanih celic z merilno skalo 50 μm. Učno množico je sestavljalo sedem, testno pa enajst slik velikosti 2088 x 1550 pikslov. Natančnost segmentacije je bila boljša pri rjavih pikslih in je v povprečju dosegla 76 odstotkov pri metriki pravilno pozitivno zaznanih pikslov. Z 51 odstotki pri isti metriki se je segmentacija modrega razreda odrezala precej slabše. The work focuses on microscopic image processing using multi-layered image processing technique for microscopic image segmentation. The purpose of the work was to optimize morphological operators for microscopic image segmentation. We have used Bayesian segmentation and a genetic algorithm for result optimisation. When learning and testing, we have used images of cells with brown and blue colouring agent with a measuring scale 50 μm. The learning set was composed of 7 images, whereas the testing set had 11 images of 2088 x 1550 pixels. The segmentation yielded better results for the brown cells with the True Positive metrics amounting to 76%. With 51% with the same metrics, the blue class did not perform as good as the brown class. genetski algoritem segmentacija celice optimizacija Bayes genetic algorithm segmentation cells optimisation Bayes true true false T. Avberšek Slovenski jezik Angleški jezik Magistrsko delo/naloga 2018-06-01 13:16:49 2018-08-24 15:46:01 2022-08-01 19:36:57 2019-07-11 18:14:28 2018 0 [Maribor 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.451.354(043.2) 21718806 URN:SI:UM:DK:HCVQT8XO MAG_Avbersek_Tomaz_2018.pdf MAG_Avbersek_Tomaz_2018.pdf 1 B4216C848B1AD154E932C8DB664EC9BA 89f55d0c26d8e84ca703c137932d8b34417d5fcb360c8bf634315aae62c1b3a3 fd9686b1-7c0f-11eb-bb7a-00155d0001ca 20.500.12556/dkum/dc142d31-5bb2-4b06-a603-dec4d1435636 https://dk.um.si/Dokument.php?lang=slv&id=125186 MAG_Avbersek_Tomaz_2018.zip MAG_Avbersek_Tomaz_2018.zip 1 8D1540B95D0DF9DF5FCAB2477F610628 a6e41f04a4fd5c3827522b16be246f8ca6ae5cf22773e9e940babaaaf7f7cfa6 fd0c1d4a-7c0f-11eb-bb7a-00155d0001ca 20.500.12556/dkum/0e24beda-37b6-40a0-8d2b-d1504bf98fef https://dk.um.si/Dokument.php?lang=slv&id=125155 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0