Research On Technologies Of Tracking And Registration For Mobile Augmented Reality

Augmented reality (AR) provides people with a new method for human-computer interaction and display of information. Though augmented reality technology, people can obtain information related to the scene more directly while interact with devices more conveniently. With the further development of hardware and software technology in the future, augmented reality will be applied in more fields and change people’s life more deeply.Tracking and registration technology is the basis for augmented reality applications. Tracking and registration mean to estimate the position and attitude of the device. After that, the generated virtual objects can be rendered in the right position, so users will feel comfortable. The purpose of this paper is to study key technologies in tracking and registration for mobile augmented reality. First, we study how to process the data of inertial measurement units (IMUs). We also study how to make better use of data of images and IMUs. Then we make a research on detecting the plane in the scene more effectively for augmented reality applications. Based on these, we study how to implement a complete real-time augmented reality system on phones. The main achievements of this paper include:(1) A research on methods that fuse data of images and IMUs was made. Based on this, an algorithm for fast feature tracking which make use of IMUs is implemented. The algorithm is suitable for tracking on phones.(2) A method for automatically detecting the plane in the scene and constructing the map is implemented for augmented reality applications. Compared with other methods, it need not people’s intervene for choosing keyframes while detects and chooses keyframes automatically. Meanwhile, with the use of data of IMUs, the method reduces the time for estimating the pose effectively.(3) A complete augmented reality system is implemented on phone platform. The system can automatically achieve initialization, tracking and registration of the device, the rendering of virtual object and relocalization. Some optimization algorithms are implemented which make use of data of IMUs, so the system can run in real-time. Lastly, two algorithms are implemented so that the system can recover it automatically when the tracking is lost.