A real-time monocular vision-based 3D mouse system

Abstract

Speed, robustness, and cost are three important factors that determine the success of a visionbased human-computer interaction system. In our system, we introduce a scheme that uses only one camera (together with a mirror) to derive the 3D coordinates of a target object. To ensure a more natural interaction between the human and the computer, our system allows the user to use his/her bare fingers to point to any position in the 3D space. In addition, robustness and speed is achieved through the use of the chain-code algorithm. The accuracy of the results is improved through the use of suitable post-processing filters, The performance of our system is thoroughly evaluated.