Outdoor marker-less tracking and pose calculating system for mobile augmented reality
The computational power of mobile Smartphone devices are ever increasing and high-end phones become more popular amongst consumers every day. Powerful processors, combined with cameras and ease of development encourage an increasing number of Augmented Reality (AR) researchers to adopt mobile Smartp...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2014
|
| Subjects: | |
| Online Access: | http://ir.unimas.my/id/eprint/8705/1/Rehman%20Ullah%20Khan%20ft.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The computational power of mobile Smartphone devices are ever increasing and high-end phones become more popular amongst consumers every day. Powerful processors, combined with cameras and ease of development encourage an increasing number of Augmented Reality (AR) researchers to adopt mobile Smartphones as AR platform. Implementation of marker-based Augmented Reality systems on mobile phones is mostly a solved problem. Marker-less systems still offer challenges due to increased processing requirements. Some researchers use mobile AR browsers but AR browsers don’t have recognition capability and can only use location data to provide information. While some adopt client server architecture for marker-less tracking which is adversely affected by network latency. Some use moveable system using laptop and much extra sensing hardware which increase the cost, unnatural in use and difficult to maintain.
This research investigates and re-designs the “speed up robust feature” (SURF) algorithm for Smartphones to enable it to recognize the real world objects. The algorithm is optimized for Smartphone by reducing the descriptor size from 64 dimensions to 32 dimensions and redesigning the algorithm into multithreaded modules. This thesis also proposes an algorithm for pose estimation using homography and investigates the use of static database in smart-phones so as to bypass the need of a desktop server. A proof of concept mobile AR system is implemented as part of this thesis. Experiments and many evaluations are conducted to validate the proposed algorithms on iPhone. The experiments show that the formulated algorithms provide stable and accurate registration, robust recognition and tracking of real world objects from natural features in a faster, easy, and more convenient way. |
|---|