Figure 8.8 - Real-time, local depth estimation, model-based augmentation
According to the MAR component classification scheme in Clause 7, this system class has the following characteristics:
Pure sensors – visual, other (3D depth)
Real world capturer – visual/video, other (3D depth)
Recognizer –3D model/primitives, recognition event, local
Tracker – 3D model/primitives, spatial event, local
Space mapper – spatial
Event mapper – location, local
Execution engine – local, 3D + t
Renderer - visual
Visual display – 3D
8.3.3 Real Time, Remote Depth Estimation, Condition-based Augmentation
Example: The end-user has an augmented reality experience where one virtual object is displayed on a horizontal plane detected within a radius of 10m.
The Device captures multi-view video, sends synchronized samples to a Processing Server that estimates the depth. This representation is sent to the device and the server uses it to detect conditions imposed by the content designer. The server sends as well the transformation matrix that the Device uses to render the virtual object by using the scale specified by the content designer. The content specified in the Information viewpoint is:
Media used for the augmentation.
The orientation and scale of the virtual object (uniform/isotropic scaling representing physical units).
The condition (e.g. horizontal plane within a ray of 10m).
URL of the Processing Server.
Figure 8.9 - Real Time, remote-depth estimation, condition-based augmentation
According to the MAR component classification scheme in Clause 7, this system class has the following characteristics:
Pure sensors – visual, other (3D depth)
Real world capturer – visual/video, other (3D depth)
Recognizer –3D primitives, recognition event, remote
Tracker – 3D primitives, spatial event, remote
Space mapper – spatial
Event mapper – location, local
Execution engine – local, 3D + t
Renderer - visual
Visual display – 3D
8.3.4 Real Time, Remote-depth Estimation, Model-based Augmentation
A content designer captures offline an approximation of the real world as a 3D model and then authors content by adding additional 3D virtual objects registered within the approximation of the real world. The end-user navigates in the real world using a multi-view camera. The captured video stream is sent to the Processing Server, which computes the depth as well as the transformation matrix of the camera in the real world. Information is sent back to the Device that uses them for augmentation. The content specified in the Information viewpoint is:
Virtual objects and their local transformations in the MAR experience.
An approximation (3D model) of the real world.
URL of the Processing Server.
Share with your friends: |
