Multi-Finger Chord for Hand-held Tablets
Multitouch-enabled technologies can detect the location and number of touch contacts with our hands. Without additional technology, however, they cannot identify the specific finger, which is touching the surface. Two-finger gestures have become pervasive in tablet gesture languages. But two fingers cannot suffice for identifying fingers reliably because they only define a segment of a line. The hand has a polygonal shape whose vertices are the fingertips. Depending on the orientation of the hand, a two-finger segment can match many segments of the hand-shape polygon. By contrast, three fingers define two segments which relative geometric properties guarantee a unique match in well- chosen cases. At least three fingers are needed for the simplest determination of the hand posture of the hand. Three fingers determine postures for a reasonably large set of noncontorted postures. The hand could curl fingers under or raise one up to increase the size of this vocabulary, but this would generate variability and decrease accuracy. We would require a more complex model of the hand to be accurate.
The Key Idea
Our hands might be of different size, but they share common hand-shape characteristics. We build a classifier that extracts relative measurements and identifies which fingers are involved into a performed multi-finger chord. Our approach is entirely based on relative measurements designed to be insensitive to variations in the actual size of users' hand. However, the minimum requirement for our classifier to be able to identify a multi-finger chord is that at least three fingers need to be held down. The classifier can distinguish 3 chord families, each containing 3 different three-finger postures that share a common concept: (1) Neighboring fingers, (2) Thumb-Pinky Basis, and (3) Thumb-Index basis.
The vocabulary |
|
Download (Note: this is the example implementation of the in the paper described classifier.)
Reference
- Julie Wagner, Eric Lecolinet, Ted Selker. "Multi-finger Chords for Hand-held Tablets: Recognizable and Memorable". [paper] (Honorable Mention)
