Pointing device based on estimation of trajectory and shape of a human hand in a monocular image sequence

摘要

A pointing device is indispensable element in composing graphical user interface. Now, mouse is widely used as a pointing device because its usage is intuitive and easy to understand. For the user to use the mouse device, the user must be able to touch the equipment directly with the provided mouse. Therefore, the area of the operation which the mouse equipment could be freely used is limited. In general, a mouse device is comprised of pointing and switch systems. If it is possible to replace the system on a nontuch device, there would be no need to consider the kind of limitation which a mouse device has. In this paper, we propose a technique to estimate the shape and position of the human hand with from a monocular camera for recognition of the human gestures. In the system, the sequential images are recorded with a monocular camera, and then transformed into the log-polar mapped images by log-polar mapping. The higher order local autocorrelation feature changes when the target is rotated. We introduced therefore a rotational invariance by carrying out the log-polar mapping before extracting the higher order local autocorrelation feature. Our system using the log-polar mapping reduces the date size of the images without spacial accuracy of the target.