The In this paper, several image processing techniques are used to achieve identifying hand signals of Rock, Paper, and Scissors. We extract the hand images from the video sequences. The skin color, and a generalized threshold value is used to map the hand image into a binary format. However, several kinds of Scissors states are difficult to judge correctly. Therefore, we use the distance between the two finger tips to improve or even get rid of these obvious error states. The segmentation, histogram, and angular criteria methods are then used to distinguish the exactly Scissors hand signal. It is the basic rule that the extended line of the hand (segment CD) shown in Fig. 8 must fall within the included angle (segment AD and BD). Finally, the scheme calculates the conditions of m BD ≥ m CD ≥ m AD and m AD ≥ m CD ≥ m BD, which are used to judge if the Scissors is correct or not. Simulated results show that our scheme is an effective method to identify the hand signals of Rock, Paper, and Scissors. (15 refs)
