Loading...
Please wait, while we are loading the content...
Similar Documents
Vision: mapping the world in 3d through first-person vision devices with mercator
| Content Provider | ACM Digital Library |
|---|---|
| Author | Dhoedt, Bart Simoens, Pieter Verbelen, Tim |
| Abstract | Many vision-based applications, especially in the domain of augmented reality, must align the camera position in the observed scene. However, traditional cameras only register textures of the observed scene. The reconstruction of depth information from 2D images is compute intensive and inevitably results in loss of accuracy. In this article, we present Mercator, a cloudlet-based system to build a 3D model of the world on which other applications can be built. The model is continuously updated, refined and expanded by crowd-sourcing depth data from 3D cameras on head-mounted devices such as Google Glass. Mercator scales up to a worldwide system by distributing the model over the network edge in geographically close cloudlets. We present the software building blocks of the system, and discuss challenges related to data management, privacy and programming model. |
| Starting Page | 3 |
| Ending Page | 8 |
| Page Count | 6 |
| File Format | |
| ISBN | 9781450320726 |
| DOI | 10.1145/2482981.2482982 |
| Language | English |
| Publisher | Association for Computing Machinery (ACM) |
| Publisher Date | 2013-06-25 |
| Publisher Place | New York |
| Access Restriction | Subscribed |
| Subject Keyword | Computer vision Crowd-sourcing 3d map Wearable computing Smartphone Cloudlet |
| Content Type | Text |
| Resource Type | Article |
