View Article |
Adaption of invariant features in image for point clouds registration
Mohd Azwan Abbas1, Halim Setan2, Zulkepli Majid3, Khairulnizam M. Idris4, Mohd Farid Mohd Ariff5, Albert, K. Chong6, Lau, Chong Luh7.
Currently, coarse registration methods for scanner are required heavy operator intervention
either before or after scanning process. There also have an automatic registration method but only applicable to a limited class of objects (e.g. straight lines and flat surfaces). This study is devoted to a search of a computationally feasible automatic coarse registration method with a broad range of applicability. Nowadays, most laser scanner systems are supplied with a camera, such that the scanned data can also be photographed. The proposed approach will exploit the invariant features detected from image to associate point cloud registration. Three types of detectors are included: scale invariant feature transform (SIFT), 2) Harris affine, and 3) maximally stable extremal regions (MSER). All detected features will transform into the laser scanner coordinate system, and their performance is measured based on the number of corresponding points. Several objects with different observation techniques were performed to evaluate the capability of proposed approach and also to evaluate the performance of selected detectors.
Affiliation:
- Universiti Teknologi MARA, Malaysia
- Universiti Teknologi Malaysia, Malaysia
- Universiti Teknologi Malaysia, Malaysia
- Universiti Teknologi Malaysia, Malaysia
- Universiti Teknologi Malaysia, Malaysia
- University of Southern Queensland, Australia
- University of Southern Queensland, Australia
Toggle translation
Download this article (This article has been downloaded 67 time(s))
|
|
Indexation |
Indexed by |
MyJurnal (2019) |
H-Index
|
0 |
Immediacy Index
|
0.000 |
Rank |
0 |
Indexed by |
Scopus (SCImago Journal Rankings 2016) |
Impact Factor
|
0 |
Rank |
Q3 (Engineering (miscellaneous)) |
Additional Information |
0.156 (SJR)R) |
|
|
|