magistrsko delo
Jakob Marolt (Author), Simon Klančnik (Mentor)

Abstract

V magistrskem delu smo reševali problem preverjanja kakovosti površin rebričenih obdelovancev z uporabo strojnega vida. Na kratko smo predstavili področje računalniškega vida s tremi algoritmi primerjanja oblik (SIFT, SURF in ORB). Izdelali smo prototipni sistem za nadzor kvalitete rebričenja s pomočjo strojnega vida. Zajete slike je procesiral računalnik Raspberry Pi model 1 B+, ki ga je operiral operacijski sistem Raspbian. Obdelovanci so bili osvetljeni z dvema belima visoko svetilnima LED diodama. Sliko je zajela standardna CMOS Raspberry Pi kamera s 5 MP. Izdelali smo računalniški program v programskem jeziku Python z uporabo standardnih modulov in knjižnice OpenCV. Primerjali smo uspešnost in čas procesiranja vseh treh algoritmov primerjanja oblik. Vsi algoritmi so 100 % uspešno ločili ustrezne obdelovance od neustreznih. Najkrajši čas procesiranja je imel program z algoritmom ORB, na drugem mestu SURF in na zadnjem mestu SIFT. Ocenili smo materialne stroške prototipnega sistema, ki znašajo 87 €.

Keywords

strojni vid;računalniški vid;preverjanje kvalitete;OpenCV;Python;rebričenje;SIFT;SURF;ORB;magistrska dela;

Data

Language: Slovenian
Year of publishing:
Typology: 2.09 - Master's Thesis
Organization: UM FS - Faculty of Mechanical Engineering
Publisher: [J. Marolt]
UDC: 004.923.021:004.93(043.2)
COBISS: 20880662 Link will open in a new window
Views: 973
Downloads: 235
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Quality control of products using computer vision
Secondary abstract: In our master thesis we inspected the quality of knurled parts with the use of machine vision. We briefly introduced the field of computer vision and explained the processes behind three popular feature matching algorithms (SIFT, SURF and ORB). We engineered a prototype system for quality control of knurled parts using machine vision. For data processing we used Raspberry Pi model 1 B+ which ran on the Raspbian debian system. We installed two white LED diodes with high brightness for lighting. Pictures were taken by a standard Raspberry Pi CMOS camera with 5 MP. The program was created in Python, using its standard modules and the OpenCV library. We analyzed success and time delay of all three feature matching algorithms. All were 100% successful in distinguishing the good parts from the bad ones. The fastest algorithm was ORB, followed by SURF and then SIFT. The material cost of the system was 87€.
Secondary keywords: machine vision;computor vision;quality control;knurling;
URN: URN:SI:UM:
Type (COBISS): Master's thesis/paper
Thesis comment: Univ. v Mariboru, Fak. za strojništvo
Pages: VI, 69 f.
ID: 10847342