Study about Natural Feature Tracking for Augmented Reality Applications on Mobile Devices
João Paulo Lima, Rafael Alves Roberto, Verônica Teichrieb, Guilherme Marques · 2015
This work aims to evaluate different ways of performing natural feature tracking on mobile devices. Two solutions have been developed. One solution is standalone, that is, uses only the mobile device, and the other solution transfers part of the processing to a server. Both solutions were implemented for the Android platform and use OpenCV, which is a computer vision library. Besides that, the standalone solution uses the Intel TBB library, which allows processing parallelization. The description of the utilized techniques as well as the details of how they were implemented is presented throughout this work. Finally, a comparative study of the techniques is presented based on the tests performed using a recent Android device.