Fast Finger Tracking System for In-air Typing Interface

Summary

We developed a system which performs 3D motion tracking of human's hand and fingers from images of a single high-frame-rate camera and that recognizes his/her typing motion in the air. Our template-matching-based method using hand textures reduces background effect and enables markerless tracking. In addition, use of a high-frame-rate camera enables recognition of rapid typing motion which is difficult to track using standard cameras. In order to realize real-time recognition, we developed hardware which parallelizes and accelerates image processing. As a result, we achieved real-time recognition of typing motion with the throughput of 138 fps (frames per second) and the latency of 29 ms.

Fig. 1 In-air Typing Interface

Fig. 2 Developed System and Recognition Results

Movie

• Demonstration [WMV, 19.2MB]

Reference

1. Kazuhiro Terajima, Takashi Komuro, Masatoshi Ishikawa: Fast Finger Tracking System for In-air Typing Interface, The 27th Annual CHI Conference on Human Factors in Computer Systems (CHI2009) (Boston, 2009.4.7) / Extended Abstracts, pp. 3739-3744. [PDF]