Face Recognition In Facial Images With Expression Change Using PCA

The face recognition is a very challenging task, it requires a combination knowledge of multi-disciplinary and multi-fields. The human face is a non-rigid body that has high similarities, when the facial expression changes, the result of face recognition will be affected dramatically. Therefore, to eliminate the facial expression change is essential in the face recognition process.The main work of this article is to solve the facial expression change problem in the face recognition process. The bit-plane method, the characteristics of block method, the wavelet transform method and the semi-face structure and odd-even component method etc. are researched to restrain the impact of expression change in the image preprocessing, feature extraction procedure. On this basis, a complete set of face recognition methods—Wavelet and Semi-face principal component analysis (WS-PCA)—is proposed. First the low-frequency components of the face image are extracted with the wavelet transformation algorithm. Then they are divided into two parts based on certain proportion and its odd-even components are calculated separately. The principal components are extracted through principal component analysis (PCA) eventually. With different weight value the distance is measured between the test facial images and sample facial images in order to find the most similar face. This method is effective to overcome the impact of the facial expression change in the face recognition process, Experimental data sheet shows that the recognition rate of this method is promoted from 80% to 96.88% compared with traditional PCA.