Research And Implementation Of The Mobile Augmented Reality Based On Android System

So far, Augmented Reality Applications on mobile phones are mostly based onthe characteristics of the sensors, e.g. GPS and the compass sensor. However, sensorslike GPS and compass are susceptible to interference, and becoming unstable. Besides,the accuracy of sensors on mobile phones is too low. So applications based on thatapproach has defect on stability and accuracy. With rapidly development of the levelof mobile phone hardware, the use of visual tracking technology is mature on mobilephones, and face tracking and recognition is the most widely used technology.Therefore, this paper focuses on the key issues of using face tracking recognitiontechnology in developing augmented reality applications on Android platform.The main research work is as follows:First, this paper does research on the face tracking algorithm that can satisfyperformance requirements of Android platform. The tracking algorithm is the firststep of augmented reality technology, the quality of the algorithm directly affect s theuser experience of augmented reality technology. The cpu and ram is limited inmobile device, this paper do research in face tracking algorithm on PC and chooseCamshift algorithm as our main algorithm. Camshift algorithm cannot track faceautomatically without initial a face area, so we initial a face area with a face detectionalgorithm. To reduce the using of resources, we narrow the search window to the sizea little bigger than the face rect size. And to improve the accuracy of the algorithm,we re-specify the search window when it make mistake. Then we test the algorithm ona Android phone, and it shows that the tracking performance of the algorithm is stableand able to meet the needs of augmented reality applications on the Android platform.Second, this paper does research on the face recognition algorithm that cansatisfy performance requirements of Android platform. Face recognition is usuallyused at the beginning, so the efficiency requirements of face recognition is lower thanface tracking algorithm, but to maintain a good user experience, face recognitiondelay cannot be too high. By doing research about face recognition algorithm usingLBP histogram, this paper gives a face recognition algorithm on Android platform.Through the use of making grayscale image into blocks and using the Gaussian filter,histogram equalization pretreatment and other methods, the accuracy and delay meetsthe needs of the Android platform. Finally, we test it on Android phones, and it shows that the algorithm can meet the expected requirements, and can be used for augmentedreality applications on the Android platform.Third, this paper does research on virtual object registration method according tothe result of face tracking and recognition. By using the method based on thecharacteristic triangular face of the estimated posture, we know the postureinformation to determine the virtual objects relative to a specific position andorientation information of the mobile phone camera, and then we draw the object onthe screen, thereby completing the registration process of the virtual object.Fourth, this paper designed and implementationed an augmented realityapplication named Augmented Face based on the above research in Android platform.The application runs well in Galaxy nexus, a smartphone using Android nativeoperating system. That shows the software designs well and verify the correctness ofthe algorithm and high-performance.