Low power gesture recognition system using skin statistics for consumer applications

Hand gestures recognition offers an alternative and natural input interface to control the consumer devices thanks to the latest advancement in the vision systems. For example, it can replace the remote control of television or set top box. However, the power consumption is a big issue in such applications and seriously affects its usability in standby modes. In this paper, we analyze if power consumption can be reduced taking advantage of the unique characteristics of such applications. We also show that skin pixel statistics are useful for designing low power systems using this framework.