Cui Wenzhe, Xu Zheer, Zhu Suwen, Yang Xing-Dong, Bi Xiaojun, Li Zhi, Ramakrishnan I V
Stony Brook University.
Dartmouth College.
Proc SIGCHI Conf Hum Factor Comput Syst. 2021 May;2021. doi: 10.1145/3411764.3445081. Epub 2021 May 7.
Back-of-device interaction is a promising approach to interacting on smartphones. In this paper, we create a back-of-device command and text input technique called BackSwipe, which allows a user to hold a smartphone with one hand, and use the index finger of the same hand to draw a word-gesture anywhere at the back of the smartphone to enter commands and text. To support BackSwipe, we propose a back-of-device word-gesture decoding algorithm which infers the keyboard location from back-of-device gestures, and adjusts the keyboard size to suit the gesture scales; the inferred keyboard is then fed back into the system for decoding. Our user study shows BackSwipe is feasible and a promising input method, especially for command input in the one-hand holding posture: users can enter commands at an average accuracy of 92% with a speed of 5.32 seconds/command. The text entry performance varies across users. The average speed is 9.58 WPM with some users at 18.83 WPM; the average word error rate is 11.04% with some users at 2.85%. Overall, BackSwipe complements the extant smartphone interaction by leveraging the back of the device as a gestural input surface.
设备背部交互是一种很有前景的智能手机交互方式。在本文中，我们创建了一种名为BackSwipe的设备背部命令和文本输入技术，它允许用户单手握住智能手机，并用同一只手的食指在智能手机背部的任何位置绘制单词手势来输入命令和文本。为了支持BackSwipe，我们提出了一种设备背部单词手势解码算法，该算法从设备背部的手势推断键盘位置，并调整键盘大小以适应手势比例；然后将推断出的键盘反馈到系统中进行解码。我们的用户研究表明，BackSwipe是可行的，并且是一种很有前景的输入方法，特别是对于单手握持姿势下的命令输入：用户可以以92%的平均准确率、5.32秒/命令的速度输入命令。文本输入性能因用户而异。平均速度为9.58字每分钟，有些用户可达18.83字每分钟；平均单词错误率为11.04%，有些用户为2.85%。总体而言，BackSwipe通过将设备背部用作手势输入表面，对现有的智能手机交互进行了补充。
