20.500.12556/DKUM-17109 PREPOZNAVA PROMETNIH ZNAKOV Z UPORABO POSTOPKOV RAČUNALNIŠKEGA VIDA RECOGNITION OF TRAFFIC SIGNS BY USING COMPUTER VISION PROCEDURES V sodobnih avtomobilih najdemo vse več pomoči vozniku: sisteme proti zdrsavanju koles, pomoč pri zaviranju, sisteme za avtomatsko parkiranje itd. Ena izmed takih pomoči je razpoznava prometnih znakov. Razpoznava prometnih znakov, ki jo ponujajo proizvajalci avtomobilov, razpoznava le omejitve hitrosti, ne pa drugih pomembnih znakov, kot so znaki za nevarnost ali znaki za izrecne odredbe. Ker so to prometni znaki, ki so pomembni, smo v tem diplomskem delu razvili preprost, a vseeno učinkovit sistem za njihovo prepoznavanje. Ta sistem temelji na segmentaciji s pragovno operacijo v različnih barvnih prostorih in iskanju osnovnih geometrijskih oblik z uporabo preprostih šablon. Rezultati kažejo, da tak pristop učinkovito prepoznava okrogle in trikotne znake, prepoznava štirikotnih znakov pa je nezanesljiva. Za uspešnejšo prepoznavanje le-teh pa je potrebnih še nekaj dodelav, ki jih v delu tudi predlagamo. There are more and more driving assistances in modern cars: traction control system, braking assistance, system for automatic parking etc. Among them you will also find traffic sign recognition. The traffic sign recognition, which is offered by car manufacturers recognizes only speed limit signs, but not other important signs, such as danger warning or prohibitory signs. In this diploma work, we implemented a simple, yet efficient system that identifies these signs aswell. The system is based on thresholding in different colour spaces and on searching for basic geometric shapes by using very simple templates. Results point out that such an approach effectivelly identifies circular and triangular traffic signs, but is unreliable at identifing rectangular traffic signs. This system still needs some modifications (which are also suggested in this work) to effectivelly identify such traffic signs. prometni znaki prepoznava obdelava digitalnih slik segmentacija računalniški vid traffic signs identification/recognition digital image processing segmentation computer vision true true false [M. Lenič] Slovenski jezik Angleški jezik Diplomsko delo/naloga 2010-12-20 22:53:10 2011-01-14 07:22:11 2022-04-13 09:12:33 2023-12-21 03:29:58 2010 0 Maribor 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.89:656.1(043.2) 14822166 URN:SI:UM:DK:PLGCWKQO UNI_Lenic_Matjaz_2010.pdf UNI_Lenic_Matjaz_2010.pdf 1 75C17B3EC8FBB1B905CF6381D69EFCB0 1b5e723d1c41a3adb4d350db008949272bc9aea39f77e9b752f4a035f2e5a789 cf41e9d3-7c04-11eb-bb7a-00155d0001ca 20.500.12556/dkum/05ddcd9a-ef84-4948-afbf-5769a00306a5 https://dk.um.si/Dokument.php?lang=slv&id=19756 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0