Study Of Gesture Recognition Technologies For Smart Glasses

As the rapid development of Head-Mount Display(HMD)and Augment Reality(AR)technology,human computer interaction(HCI)is getting more and more important.Traditional human computer interaction system based on keyboard and mouse can't satisfy the demand.Gesture recognition technology makes it possible for users to interact with virtual objects displayed on HMD using their bare hands in the AR scene.Therefore,gesture recognition catches more and more attention as the key technology of next generation human computer interaction system.The research on gesture recognition technology is at early stage in laboratory.Most algorithms of gesture recognition can only run on desktop computers in laboratory.In practice,wearable devices can't support gesture recognition due to limitation of power and size.On the other hand,gesture cognition technology which can run on wearable devices is inaccuracy and slow.This work designs and implements a hand gesture recognition system based on smart glasses.This paper combines depth information and color information,proposes an adaptive skin color model based on Haar-like features detection,which improves the accuracy of gesture recognition under different illumination conditions.What's more,This paper proposes a fast hand center estimation algorithm,improving the speed of the hand gesture recognition system.This work combines hand gesture recognition system specially designed for wearable device and 3D display technology based on Unity.This paper simultaneously executes gesture recognition and 3D display on Android-based smart glasses using multi-thread technique,and develops a 3D interaction AR application.