Scene recognition with bag of visual nouns and prepositions
The loop closure problem is central to topological simultaneous localization and mapping (SLAM); by associating features between distant portions of a trajectory, the odometry error that has accumulated between two observations can be eliminated and a more consistent map can be built. Bayesian pattern recognition techniques such as bag of visual words (BoVW) have recently shown outstanding results in solving the loop closure problem completely in image space using very simple, inexpensive cameras, without the requirement for highly accurate metric information, 3D reconstruction, or camera calibration. In this paper, a modified BoVW descriptor that incorporates simple geometric relationships within an image is used with the fast appearance-based mapping (FAB-MAP) algorithm. In direct comparisons with the traditional BoVW descriptor, an improved recall rate is observed with an acceptable increase in computational time. The proposal of a BoVW-compatible descriptor and the use of the proposed descriptor with a well-known BoVW classifier demonstrate the ability of the BoVW metaphor to be generalized, which could pave the way for more various BoVW descriptors in the same way that many individual visual feature descriptors exist within the computer vision community.
KeywordsBag of visual words Scene recognition Loop closure Place recognition SLAM
This research was supported by the MOTIE under the Industrial Foundation Technology Development Program supervised by KEIT (No. 10051155) and by Basic Science Research Program through the NRF funded by MSIP (No. 2007-0056094).
- 1.Thrun S (2001) Probabilistic robotics. MIT Press, CambridgeGoogle Scholar
- 2.Bradski G (2000) OpenCV. Dr Dobb’s J Softw ToolsGoogle Scholar
- 7.Pérez J, Caballero F, Merino L (2015) Enhanced Monte Carlo localization with visual place recognition for robust robot localization. J Intell Robot Syst 1–16. doi: 10.1007/s10846-015-0198-y
- 8.Yang C, Shengnan C, Jingdong W, Quan L (2014) Low-rank sift: an affine invariant feature for place recognition. Comput Res Repos 1–5. arXiv:1408.1688
- 9.Sünderhauf N, Dayoub F, Shirazi S, Upcroft B, Milford M (2015) On the performance of ConvNet features for place recognition. Comput Res Repos 1–8. arXiv:1501.04158
- 10.Cao J, Chen T, Fan J (2014) Fast online learning algorithm for landmark recognition based on BoW framework. IEEE Trans Ind Appl 1163–1168. doi: 10.1109/ICIEA.2014.6931341
- 11.Johns E, Yang G (2014) Pairwise probabilistic voting: fast place recognition without RANSAC. Comput Vis ECCV 505–519. doi: 10.1007/978-3-319-10605-2_33
- 13.Duda R, Hart P, Stork D (2000) Pattern classification. Wiley, New YorkGoogle Scholar
- 14.Bay H, Tuytelaars T, Van Gool L (2006) SURF: speeded up robust features. Comput Vis ECCV 404–417. doi: 10.1007/11744023_32
- 16.Rublee E, Rabaud V (2011) ORB: an efficient alternative to SIFT or SURF. Comput Vis ECCV 2564–2571. doi: 10.1109/ICCV.2011.6126544
- 17.Calonder M, Lepetit V, Strecha C, Fua P (2010) Brief: binary robust independent elementary features. Comput Vis ECCV IV:778–792. doi: 10.1007/978-3-642-15561-1_56
- 18.Cormen T, Leiserson C, Rivest R, Stein C (2001) Introduction to algorithms, 2nd edn. MIT Press, CambridgeGoogle Scholar