Research In Mobile Augmented Reality System Based On Natural Feature Points

Augmented Reality (AR) as virtual reality technology expansion, it has been the rapid development in various fields and show a strong value in the last ten years. With the popularity of mobile intelligent terminal, mobile Augmented Reality has gradually become a research branch of virtual reality technology field, and have great development potential in the industry, products, medical research, education, so it has a broad application prospects. This study mainly focused on Mobile Augmented Reality registration and system implementation based on nature feature point, the main contents as follows:1. Described in detail about Augmented Reality. Mobile Augmented Reality direction of development and applications, also analyzes the key technology in Augmented Reality tracking registered and current research progress.2. Described the natural feature points-based Mobile Augmented Reality workflow, analysis natural feature point extraction process in the workflow, demonstrate the principle of Augmented Reality3D registration, included the introduction of the coordinate system in the computer vision, perspective transformation the homography matrix concept. Finally, expounded the fusion technology.3. For Mobile Augmented Reality performance and functional requirements, two commonly used algorithms FAST and SURF are analyzed in detail and research. On this basis, proposed the strategy of combining non-scale space detection algorithm FAST and scale space algorithm SURF. The experimental results show that this detection method can satisfy the mobile Augmented Reality real-time and accuracy.4. Study and research in Augmented Reality’s feature points matching and tracking, use KD tree to complete the storage and matching of feature points, combine LK optical flow algorithm to achieve real-time tracking of feature points, whereas LK optical flow algorithm is affected by light conditions, propose an improved LK optical flow algorithm combined tracking feature point detection method.5. Relying on image and graphics toolkit and computer vision library, completed the corresponding detection matching, tracking module, combined to a simple Mobile Augmented Reality system based on natural feature points. This system use3DS file import to generate complex3D scene model.Experimental results show that the system can be completed in a real environment to enhance the image display, and satisfy the theMmobile Augmented Reality real-time requirements.