| 摘要 |
PURPOSE: A method of detecting a face intelligently regardless of the size and the rotation is provided to separate the estimation of the rotation and the identification of the face, thereby speeding up the process and improving the accuracy. CONSTITUTION: In the method of detecting a face intelligently, in step one, a still image is received through an input device such as a camera. In step two, a roughly estimated area of the face is extracted from the background image using differential imaging method. In step three, a hierarchical pyramid image is made. In step four, The pyramid image is scanned with 20 by 20 window applied. In step five, a histogram equalization is conducted to adjust the brightness. In step six, the rotated angle of the face is estimated. In step seven, an up-front face is recovered using the estimated rotation angle. In step eight, the recovered image is compared with the input image.
|
