| 摘要 |
PURPOSE:To provide a pointed position detecting method which is lightened in operational burden while an area to be processed for information input is not limited to the illustrated visual field area of a TV camera. CONSTITUTION:Information 18 on the longitudinal and lateral dimensions of a reference body is previously inputted to a computer 12. Then, a processed image is fetched by the TV camera 11 and a reference body image is extracted. Then, the parts of the four corners of the extracted reference body image are recognized to calculate their coordinate values. Then, the area of a hand is extracted from the processed image. A pointed part is recognized from the image of the extracted hand and its coordinate values (X, Y) are calculated. Lastly, normalized position information 20 on the finger tip part is calculated by utilizing the coordinate values of the four corners which are thus detected, the coordinate values of the pointed part, and the reference body information 18. |
