Group Touch: Distinguishing Tabletop Users in Group Settings via Statistical Modeling of Touch Pairs

論文URL:http://dl.acm.org/citation.cfm?doid=3025453.3025793

論文アブストラクト:We present Group Touch, a method for distinguishing among multiple users simultaneously interacting with a tabletop computer using only the touch information supplied by the device. Rather than tracking individual users for the duration of an activity, Group Touch distinguishes users from each other by modeling whether an interaction with the tabletop corresponds to either: (1) a new user, or (2) a change in users currently interacting with the tabletop. This reframing of the challenge as distinguishing users rather than tracking and identifying them allows Group Touch to support multi-user collaboration in real-world settings without custom instrumentation. Specifically, Group Touch examines pairs of touches and uses the difference in orientation, distance, and time between two touches to determine whether the same person performed both touches in the pair. Validated with field data from high-school students in a classroom setting, Group Touch distinguishes among users "in the wild" with a mean accuracy of 92.92% (SD=3.94%). Group Touch can imbue collaborative touch applications in real-world settings with the ability to distinguish among multiple users.

日本語のまとめ:

(指の方向がとれる)タッチパネルへの入力が、同一ユーザーによるものか、他のユーザーによるものかを識別する手法。2つのタッチ入力から3つの特徴量(指の相対的な角度、距離、タッチの時間差)を抽出し分類する。

(101文字)

発表スライド: