Summary: | 碩士 === 南台科技大學 === 電機工程系 === 98 === This thesis incorporates Principal Component Analysis and Evolutionary Support Vector Machine for real-time face recognition. The proposed system consists of the face detection and the face recognition subsystems. In the face detection subsystem, it uses the facial color filter to partition the possible facial color areas, and then applies the connected component labeling procedures to localize the objective faces. Afterward, the system will extract the face from the image, and then normalize these faces blocks for the later face recognition procedures. In order to recognize the faces, the principal component analysis method is adopted to determine the eigen structure of each face. After constructing the database of the weight vectors for all sample faces, we utilize coordinate transformation to obtain the 2D projections of the faces, and classify those with a criterion based on the evolutionary support vector machine theory. In the experimental results, the proposed real-time face recognition system has shown faster and more precise recognition abilities.
|