الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 108 |  |  |
| | | from | 108 | | |
Abstract
Recently, augmented reality technology has become more stable and integrated into daytime applications. Augmented reality uses tracking techniques to capture the environmental features. Tracking is classified into two types: outdoor and indoor. Currently, outdoor tracking is popular for outdoor navigation applications using the Global Positioning System (GPS). However, GPS has low performance in indoor tracking owing to the imprecision of GPS satellite signals. Indoor tracking provides a solution for the vast and complex navigation of indoor environments. This thesis provides a detailed survey of the various indoor Augmented Reality (AR) tracking techniques that have been reported in literature, along with their respective strengths and weaknesses, providing readers with a detailed survey of the current state of the field. This thesis also introduces an indoor tracking model that combines capabilities of smartphone sensors data and computer vision technologies. The presented system utilizes the oriented Features from Accelerated Segment Test (FAST) with Binary Robust Independent Elementary Features (BRIEF) to apply Oriented FAST And Rotated BRIEF (ORB) algorithm for feature extraction. The Brute Force Match (BFM) uses with K-nearest neighbor (KNN) for matching, This yields high accuracy in reaching destinations in a more effective way than previous systems. The proposed system also employs the A* algorithm to determine the shortest path and cloud computing to save our database.
Abstract
vi
One of the significant advantages of cloud computing in our model is the ability to save and store data in the cloud, rather than relying on local storage devices such as hard drives or on-premises servers. The model enables the users to select a destination without predefined maps and to simultaneously calculate the shortest path. Experiments demonstrate that the proposed model achieves an average accuracy of approximately 99 percent within a 7-10 cm error bound in scenarios involving different distance paths, in the range of 20 to 80 meters. The experiments also showed that all users successfully reached their destination. The error of the proposed model was significantly lower than the errors reported in the literature for research conducted with markerless technology and tested in diffirent area sizes.