xstroke: Full-screen Gesture Recognition for X

Carl D. Worth
Information Sciences Institute, USC


Gesture recognition is a common method of text input on handheld and other pen-based computing devices. Xstroke is a full-screen gesture recognition program for the X Window System.

The touchscreen of a typical pen-based device is divided into two regions, a primary region for application display and interaction, and a secondary region for gesture input. Full-screen gesture recognition improves on the typical implementation by sharing the entire screen for both purposes. Applications benefit as more screen real estate is available. Recognition performance improves due to increased information from larger gestures. Usability increases as less time is lost switching attention between two separate screen regions.

Full-screen gesture recognition presents several user-interface design challenges. Conventional GUI interaction is pointer driven. This makes it difficult for systems with single-button pointer devices to distinguish between GUI interaction and character input. Several solutions to this problem are examined.

The simple feature-based recognition engine at the heart of Xstroke has proven capable of resolving gesture sets of 100 different gestures with up to 95% accuracy. Xstroke is freely available and has become the predominant gesture recognition system for X-based handheld devices.

Full text:

Presented at the 2003 Usenix Annual Technical Conference in San Antonio, Texas, June 2003.

Carl Worth
Last modified: Mon Jul 28 22:07:36 EDT 2003