| Summary: | 碩士 === 國立成功大學 === 電機工程學系 === 102 === In recent years, fast processing of image recognition is required for embedded systems, e.g., embedded automotive systems, security systems, or portable device. Face detection is the most important step in face recognition systems with applications to face tracking and recognition. Unfortunately, when frame size increase, the speed of face detection will extremely decrease and false positive rate will increase. In this thesis, we implement the rapid and high correct rate face detection method based on Adaboost algorithm. Also, we choose YCbCr color space to filter pixels of the entire image first and to detect the faces of image. By this way, we can remove the most parts of the image which don’t need to be detected in order to reduce the false detection rate.
In addition, the hardware architecture for face detection system is implemented.
Experimental results show that the proposed architecture can achieve high detection rate (80.5 %) at 22 frame/second for image size of 640 × 480 pixels.
|
|---|
