John Robinson's pages on
Research

INTRODUCTION
Visual Info Engineering

IMAGE CODING
Quincunx / BTPC
Object-based coding
Error resilience
Primitive-based coding

IMAGING HUMANS
Face feature space
Fast face tracking
Facial image coding
Facial image enhancement

AUGMENTED REALITY
Wearables
Perspective registration
Video Augmentation
The WristCam

RECENT PUBLICATION
Patents
Journals
Conferences

RESOURCES
Media Tech Resources

RETURN TO:
Userport homepage
York homepage

The Wrist Cam

We are investigating the use of a wrist- or forearm-mounted camera for wearable computer applications. The WristCam attaches to the underside of the forearm and views only the front of the hand and fingers. Typically it is invisible to other people, being concealed by the wearer's sleeve and connected up the sleeve to the remainder of the system. Foreseen applications for the WristCam include a hand's-eye-view key frame recorder (useful for storing sequences of gestures like keystrokes on a dial pad), hand scanning of documents, and active vision tasks that can combine input from a static head camera and a dynamic wrist camera. However, our main interest is to develop a virtual input device that allows chorded character input by tracking certain coarse finger movements or gestures. To do this, we track and interpret finger motions as seen by the WristCam, in the wide range of illumination conditions and backgrounds encountered as the user moves their hand around. Typically, the user taps fingers against a handy table top or their clothing, or simply moves their fingers freely. Recognizing these motions and converting them efficiently to system input makes it possible for WristCam users to interact with their wearable computers unobtrusively. Although head-mounted displays are becoming less encumbering and obvious to others (non-wearers), existing input devices such as chord keypads are very visible and distracting to others. The WristCam allows input to be subtle (or even covert) and therefore not distracting. Also, when the system is not in "input" mode the user's hands are instantly free for other uses.

Image by Andrew Vardy