Real-time tracking of multiple fingertips and gesture recognition for augmented desk interface systems

We propose a fast and robust method for tracking a user's hand and multiple fingertips; we then demonstrate gesture recognition based on measured fingertip trajectories for augmented desk interface systems. Our tracking method is capable of tracking multiple fingertips in a reliable manner even in a complex background under a dynamically changing lighting condition without any markers. First, based on its geometrical features, the location of each fingertip is located in each input infrared image frame. Then, correspondences of detected fingertips between successive image frames are determined based on a prediction technique. Our gesture recognition system is particularly advantageous for human-computer interaction (HCI) in that users can achieve interactions based on symbolic gestures at the same time that they perform direct manipulation with their own hands and fingers. The effectiveness of our proposed method has been successfully demonstrated via a number of experiments.