TunnelSlice

Also known as Egocentric tunnel, Procedural slicing

In wearable AR the user frames a region with a two-handed pinch to cast a tunnel into the scene, then slices off the near and far ends to leave a box around the objects they want.

How it works

Selects a 3D subspace (a cuboid) rather than single objects. In the targeting step the user drags a rectangle on the view with pinch-tip pointers of both hands; the 2D frame is projected along the view direction into a tunnel, and the tunnel is automatically trimmed to the depth range of the objects it collides with. In the refinement step further two-handed pinches move the near and far planes to cut away unwanted objects in front of or behind the region. Tested with 15 participants on a video see-through RGB-D head-worn display across six scenarios varying occlusion and whether an object sat in the centre; compared with CubeSelect and an encapsulating bounding box, it was preferred, more stable across scenarios, and best when occlusion was strong without a central object.

Papers

Hyeongmook Lee, Seung-Tak Noh, Woontack Woo (2017). TunnelSlice: Freehand Subspace Acquisition Using an Egocentric Tunnel for Wearable Augmented Reality. IEEE Transactions on Human-Machine Systems 47(1), pp. 128–139 doi:10.1109/THMS.2016.2611821