Invariant 3D line context feature for instance matching
- Authors
- Cho, K.[Cho, K.]; Kim, J.[Kim, J.]; Lee, S.[Lee, S.]
- Issue Date
- 2019
- Publisher
- Springer Verlag
- Keywords
- 3D line matching; Directional/locational contexts; Multiple solutions or interpretations
- Citation
- Advances in Intelligent Systems and Computing, v.935, pp.473 - 485
- Indexed
- SCOPUS
- Journal Title
- Advances in Intelligent Systems and Computing
- Volume
- 935
- Start Page
- 473
- End Page
- 485
- URI
- https://scholarworks.bwise.kr/skku/handle/2021.sw.skku/15193
- DOI
- 10.1007/978-3-030-19063-7_37
- ISSN
- 2194-5357
- Abstract
- Conventional Approaches to line segment matching have shown their performances less satisfactory mainly since some of the features used for matching, such as the center and the starting/ending points of line segments, are not invariant. Furthermore, a pair of line segment sets to be matched may not have one to one correspondence, but each can be a subset of the other. This led to multiple solutions or interpretations in matching, where finding out all the possible solutions or interpretations out of an arbitrarily overlapping pair of line segment is of an issue. This paper presents a general method of identifying all the possible solutions or interpretations for an arbitrary pair of line segment sets by using invariant features associated with line segments. The invariant property of line segments comes from the orientation and location contexts of line segments that are defined based on infinite line representation of individual line segments. Simulation and experiment shown the effectiveness of the proposed method compared to conventional methods. © Springer Nature Switzerland AG 2019.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - Information and Communication Engineering > School of Electronic and Electrical Engineering > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/skku/handle/2021.sw.skku/15193)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.