Research Of Augmented Reality System Based On Mobile Platform

Augmented reality is the technology that the virtual information can be applied tothe real world through the computer, so that the real environment and virtual objects inreal-time is superposed onto one screen or space exists at the same time. In recent years,Augmented reality system based on mobile platform has been widely studied owing totheir broad application prospects. Due to their several advantages such as positiveuser-experience and portability, AR on mobile platform has been widely used in allaspects of our life. This dissertation is mainly concerned with the implementation ofsome particularly crucial techniques. The author’s major contributions are outlined asfollows:1. Technology of registration has been studied. The main work is divided into twoparts: Firstly, Registration based on artificial markers was achieved under theframework of ARToolkit. Secondly, Registration based on natural markers. First of all,this work was implemented by the Speeded Up Robust Features(SURF) algorithm.Followed, Against the SURF which cannot meet the requirements of real-timeprocessing, an algorithm on the basis of feature training for fast feature matching isintroduced. The results trained from the image library, which is built by this algorithmitself will be saved as bit. So that feature matching can be achieved by taking advantageof bitwise operation. Because of the using of bitwise operation, the algorithm takes atiny fraction of processing time, and in the latter case a fraction of the memory as well.Therefore, this algorithm is very suitable for mobile platforms which have limitedprocessing power to realize feature matching.2. Display technology also has been studied. Display technology is implementedon the openGL library, and it achieves the goal that loaded one bitmap into the marker.Firstly, it converts the bitmap to texture data; Then the texture data is loaded on themarker coordinates of the marker in the video; So the display technology for augmentedreality can be achieved.3. The system used to recognize natural markers on mobile platform has beenachieved. An algorithm on the basis of feature training for fast feature matching wasused in this system. The work of natural markers recognition is an important foundationto the augmented reality system which is based on natural marker.