| 摘要 |
<p>A face (skin color) area of a person is extracted at a minimized error irrespective of different skin colors of the human races. A hue value calculating portion (1) determines a hue value from an input image (RGB signal). A primary discriminating portion (2) extracts pixels whose hue value lies in a specified range defined by limiting values outputted from a control portion (5). A pixel-counting portion (3) counts the extracted pixels. The control portion (5) selects a threshold value for extracting a face area according to the count of the extracted pixels and outputs the threshold value to a face-area extracting portion (4) which in turn extracts a face area according to the threshold value. The method and device are usable for video processing device, e.g. for producing video information suitable to use in video telephones and video conferences.</p> |
