Abstract
dc:descriptionImage Matching Basis of computer vision - image matching and recognition, the beginning of which is associated with 1981. Hans P. Moravec corner detection algorithm. Some of the first recognition algorithms were called - SIFT and SURF. These were patented, and an alternative for them was created - ORB. In this work, mentioned algorithms were analyzed. Analysis of algorithms were performed by comparing training images with images with such transformations as scale, rotation and illumination. A total of 107 tests with photos were made. According to the results obtained, it can be stated that the best results in all cases were shown by the SIFT algorithm, although the speed was the lowest. For performance comparison effectiveness was calculated. Highest effectivness had - SURF, SIFT the second, due to large number of keypoints detected in photos, many of which had too little information for later matches. The poorest results were achieved using ORB algorithm. For that impact also had matches located arround the image center while SIFT and SURF evenly distributed. However, ORB showed the highest computational speed. If I had to choose for usage one of these algorithms, definetely would be SIFT.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2018
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Naina, Tomas,
- Contributors dc:contributor
-
- Beniušė, Margarita
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD25856227&prefLang=en_US