Please use this identifier to cite or link to this item:
https://scholarbank.nus.edu.sg/handle/10635/23152
DC Field | Value | |
---|---|---|
dc.title | An effective scene recognition strategy for biomimetic robotic navigation | |
dc.contributor.author | TEO CHING LIK | |
dc.date.accessioned | 2011-06-10T18:03:09Z | |
dc.date.available | 2011-06-10T18:03:09Z | |
dc.date.issued | 2007-05-15 | |
dc.identifier.citation | TEO CHING LIK (2007-05-15). An effective scene recognition strategy for biomimetic robotic navigation. ScholarBank@NUS Repository. | |
dc.identifier.uri | http://scholarbank.nus.edu.sg/handle/10635/23152 | |
dc.description.abstract | Scene recognition is an important component of biomimetic navigation. In this thesis, a novel Scene Recognition Strategy (SRS), inspired from computational models of human visual saliency and a special form of motion performed by certain flying hymenopterans known as the Turn-Back-and-Look (TBL) flight, is proposed. The proposed SRS is tolerant to various forms of image distortions - viewpoint changes, illumination changes and changes in scene content due to the dynamic nature of the environment. To this end, the spatial configuration of the detected salient landmarks is encoded over the HSV colour space using SURF keypoints to provide a viewpoint and illumination invariant representation of the scene. A robust scene decision module that exploits ordinal constraints in computing a novel measure of scene similarity, together with an adaptive decision threshold, is also proposed. Experimental results show that the proposed SRS is accurate even for challenging scenes in both indoor and outdoor environments. | |
dc.language.iso | en | |
dc.subject | Biomimetics, robotic navigation, place recognition, robust data association, SLAM loop closing, Visual saliency | |
dc.type | Thesis | |
dc.contributor.department | ELECTRICAL & COMPUTER ENGINEERING | |
dc.contributor.supervisor | CHEONG LOONG FAH | |
dc.description.degree | Master's | |
dc.description.degreeconferred | MASTER OF ENGINEERING | |
dc.identifier.isiut | NOT_IN_WOS | |
Appears in Collections: | Master's Theses (Open) |
Show simple item record
Files in This Item:
File | Description | Size | Format | Access Settings | Version | |
---|---|---|---|---|---|---|
MENG_thesis_amended.pdf | 12.38 MB | Adobe PDF | OPEN | None | View/Download |
Google ScholarTM
Check
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.