| Summary: | 碩士 === 遠東科技大學 === 電機工程研究所 === 100 === According to the rapid development of image processing techniques, face detection is increasingly used in identification and diagnostic tools for medical rehabilitation. As the result, facial recognition technology has received more attention in academia and business in recent years. Moreover, research and development of this technology has been significantly increased.
This thesis, an experimental study of face detection algorithm and the angle tracking system will been proposed. Based on these color spaces, the human face and calculate the angle of the face deflection can been accurately detected. Firstly, the human face images are captured using a camera. Secondly, using the color space conversion technology and the binary method, the skin color in the color image is isolated. The use of morphological filtering of erosion and dilation are to remove noises. Thirdly, the largest portion of the binary image is then identified using connected-component labeling. Finally, we use Hough transform to find the position of the two eyes by detecting the ellipse. In the face horizontal angle detection part, we use curve fitting to identify the best fitted curve between the inputs and outputs. This ensures that the results of the detection angle are as accurate as possible.
|
|---|
